,___ ___ ___ AD-A248 709 -AL-TP-1992-0007 lllllillllllllllir

COMPARATIVE ANATOMY OF MAINTENANCETASKS (CAMT): A FEASIBIUTY STUDY

A Andrew P. Chenzoff DTICR ELECTE

Applied Science Associates, Incorporated (ASA) APR 1 4 1992M P.O. Box 1072, 292 Three Degree Road

S Butler, PA 16003 C

TR Donald R. Loose, Captain, USAF

0 Center for Supportability and Technology InsertionAir Force Logistics Command

G Wright-Patterson Air Force Base, OH 45433-5000

HUMAN RESOURCES DIRECTORATEL LOGISTICS RESEARCH DIVISION

A Wright-Patterson Air Force Base, OH 45433-6503

B0R March 1992

A Final Technical Paper for Period 1 September 1989 - 1 September 1990

T0Ry Approved for public release; distribution is unlimited.

92-0949492 4 13 087 AMiNE 11

AIR FORCE SYSTEMS COMMAND IIBROOKS AIR FORCE BASE, TEXAS 78235-5000

NOTICES

This technical paper is published as received and has not been edited by thetechnical editing staff of the Armstrong Laboratory.

When Government drawings, specifications, or other data are used for any purposeother than in connection with a definitely Government-related procurement, the UnitedStates Government incurs no responsibility or any obligation whatsoever. The fact thatthe Government may have formulated or in any way supplied the said drawings,specifications, or other data, is not to be regarded by implication, or otherwise in anymanner construed, as licensing the holder, or any other person or corporation; or asconveying any rights or permission to manufacture, use, or sell any patented inventionthat may in any way be related thereto.

The Office of Public Affairs has reviewed this paper, and it is releasable to theNational Technical Information Service, where it will be available to the general public,including foreign nationals.

This paper has been reviewed and is approved for publication.

DONALD R. LOOSE, CAPT, USAF BERTRAM W. CREAM, ChiefTask Monitor Logistics Research Division

Form ApprovedREPORT DOCUMENTATION PAGE OMB N. 0704-0188

Public eorting burden for' thi coection ofrl omeilon ks elmaoed to aIfoe r response. including the time for r Inbu e . eechng e ising ecee.dmelh ing the det needed, and coml etlng end r iewing the collection of infonntion Send comments regarding Ot burden estimate or ay other m d of tis c ol

11ronall u. inn sugeetone for reducing this burden, to Washington eadquters Serices, 0irectorls tor Wormeation Operetkns and Rempoa, 1215 Jeferson Doas Highw". Suds1204, Adington. VA a22 2.W402. end to the Office ot Menegement end Budget Peperwork Reduction Project (0704-0181). Weshtington, DC 20503.

1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED

March 1992 Final - 1 September 1989 - 1 September 19904. TITLE AND SUBTITLE S. FUNDING NUMBERS

Comparative Anatomy of Maintenance Tasks (CAMT): A Feasibility StudyC - F33615-87-D-0661PE - 62205F

6. AUTHOR(S) PR - 1710Andrew P. Chenzoff TA - 00Donald R. Loosa WU - 53

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) L. PERFORMING ORGANIZATION

REPORT NUMBER

Applied Science Associates, Incorporated (ASA)P.O. Box 1072, 292 Three Degree RoadButler, PA 16003

9. SPONSORING/MONITORING AGENCY NAMES(S) AND ADDRESS(ES) 10. SPONSORINGMONITORING AGENCYArmstrong Laboratory REPORT NUMBERHuman Resources Directorate AL-TP-1 992-0007Logistics Research DivisionWright-Patterson Air Force Base, OH 45433-6503

11. SUPPLEMENTARY NOTES

Armstrong Laboratory Technical Monitor: Captain Donald R. Loose, (513) 255-3871

12a. DISTRIBUTIONIAVAILABILITY STATEMENT 112b. DISTRIBUTION CODE

Approved for public release; distribution is unlimited.

13.ABSTRACi (Maximum 200 words)

During the system engineering process, how does one predict the maintenance Manpower, Personnel,Training, and Safety (MPT&S) consequences of a proposed design before it materializes as a breadboardor mockup? This study proposes the use of a baseline comparison with an integral data base that is moresophisticated and powerful than currently available approaches. CAMT would define a set of componenttask primitives, each common to many maintenance tasks, associated with a common set of maintainer skilland comprehension requirements. Associated with each primitive would be immediately accessible dataconcerning typical MPT&S resources required for adequate primitive performance on a variety of tasks andsystems. Methods for task primitive definition were developed, then taken into the field for testing. Taskprimitives were defined to cover three remove-and-replace tasks in one aircraft maintenance field (engines).A literature review showed this approach to be unique. Future studies will address electronic maintenance,data base acquisition, and reliability of the methodology.

14. SUBJECT TERMS 1. NUMBER OF PAGESDesign for maintainability Manpower System engineering 114Human engineering Personnel TrainingHuman-systems integration Safety 16. PRICE COE

17. SECURITY CLASSIFICATION 11. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT

OF REPORT OF THIS PAGE OF AUTLCTUnclassified Unclassified assifed UL

NSN Th 0o0.1-560 Standard Form2

i 298-102

TABLE OF CONTENTS

Section Page

I. THEORY 1-1

The Requirement 1-1

Associated Concepts 1-4

Comparative Anatomy 1-4The Scenario 1-5Traditional Discipline Boundaries 1-5Hierarchy of Design Influence 1-8Human Performance Prediction 1-10Current Manpower/Personnel/Training Practice

and Its Limitations 1-12

The CANT Concept 1-13

The CANT Task Primitive 1-13The CAMT Data Base 1-15

Structuring the CANT Data Base: An Example 1-18

Removal and Replacement of the F-16 Jet Fuel Starter 1-18Errors and Error Consequences 1-20Design-Specific Characteristics and Constraints 1-20Skill and Comprehension Requirements 1-22

Human-Systems Integration Analysis With CANT 1-24

CANT Applicability 1-30

First Steps 1-30

II. FIELD STUDY 2-1

Data Collection 2-1

Data Analysis 2-5

Results 2-6

Generality 2-6Orthogonality 2-6Completeness 2-6

Discussion 2-10

iii

TABLE OF CONTENTS

Section Page

III. ANNOTATED BIBLIOGRAPHY 3-1

Introduction 3-1

Bibliography 3-1

Summary of Literature Review 3-28

IV. SUMARY AND CONCLUSIONS 4-1

V. REFERENCES 5-1

APPENDIX A: BEHAVIOR OBJECTS, CONSTRAINTS, AND ERRORS FOR THREETASKS ON THREE AIRCRAFT A-1

APPENDIX B: LIST OF ACRONYMS B-i

LIST OF FIGURES

Figure Page

1. System Engineering Hierarchy of Design Influence 1-9

2. Human Performance Hierarchy 1-11

3. Overview: CAMT Analysis 1-25

4. Comparative Anatomy of Maintenance Tasks Focus 1-30

5. First and Second Draft, Task Primitives 2-7

6. Comprehension and Skill Requirements for ThreeTask Primitives 2-8

7. Assignment of Task Primitives to Tasks 2-9

iv

Preface

The Armstrong Laboratory Logistics Research Division (ALA() fundedthis effort via Task Order No. 5 of Air Force Contract F33615-87-D-0661.Applied Science Associates, Inc. (ASA) under subcontract to QuesTech ResearchDivision (QTRD) of QuesTech, Inc. collected, analyzed, and presented theinformation in this report. QTRD, as the prime contractor, provided TotalQuality Management of the product. AL/HRG provided final editing. Dr. AndrewP. Chenzoff of ASM served as Principal Investigator. David Shipton of ASAmanaged the field data collection. Mr. Arthur Steczkowski of QTRD served asTask Monitor. Capt Donald R. Loose of Acquisition Logistics Divisionconceived and managed the program.

Acs.ftlaa For

byv

AillylDst Spe*1alLit

__ ' .4pele J

SUMMARY

During the system engineering process, how does one predict themaintenance Manpower, Personnel, Training, and Safety (MPT&S) consequences ofa proposed design before it materializes as a breadboard or mockup? Thisstudy proposes the us* of a baseline comparison with an integral data basethat is more sophisticated and powerful than currently available approaches.

The purpose of this study was to explore the basis for defining a set ofcomponent task primitives, each common to many maintenance tasks, associatedwith a common set of requisite maintainer comprehension and skill require-ments. These would be defined midway between whole maintenance tasks (atwhich level MPT&S analysis typically occurs) and detailed task steps (at whichlevel human engineering analysis typically occurs). Associated with eachprimitive would be immediately accessible data concerning typical MPT&Sresources required for adequate primitive performance on a variety of tasksand systems.

The project has accomplished a selective review of related military andcivilian literature to:

1. determine whether an approach similar to CANT has been previouslyemployed,

2. refine the CAMT concept and its proposed application, and

3. benefit from related experience of previous workers in the field.

Methods for task primitive definition and associated comprehension/skillrequirements identification were developed (Phase 1). These methods weretaken into the field to explore the feasibility of defining task primitivesfor three remove-and-replace tasks in one aircraft maintenance field (enginemaintenance) and to test whether task primitives developed for one weaponsystem could have applicability to other weapon systems (Phase 2). This isthe first of a series of studies scoping the risks, costs, and benefits ofdeveloping a comprehensive human performance data base for use during weaponsystem acquisition.

vi

SECTION I. THEORY

The Requirement

Within the Weapon System Acquisition Processes (WSAP) there traditionallyexists a cultural rift between engineering and Integrated Logistics Support(ILS) functions. One manifestation of this rift is the lack of an integratedapproach to efficiently incorporate human performance into weapon systemperformance.

Engineers (including human engineers) create and configure weaponsubsystems and components to meet technical specifications. Meanwhile, ILSplanners (including manpower, Personnel, and Training (MPT) planners) mustadjust the military "infrastructure" to provide sufficient (human) resourcesto support the new weapon, once fielded. The engineers completely control asmall piece of the physical universe while the logistics planners work withinlarge bureacracies of people and materiel. Not suprisingly, the twocommunities of experts typically have different training, career progressions,analysis tools, and jargon. They perform their functions separately.

Logisticians traditionally perform two functions in the WSAP. First,during early weapon system requirements specification, they levy hardinfrastructure constraints within which the weapon system must function (e.g.,the number of levels of maintenance, compatibility with existing groundsupport equipment, or the requirement to be repairable with common handtools). Second, as engineers evolve supportability requirements duringtradeoff studies and system design, logisticians provide advanced planning tomeet these requirements (e.g., ordering long-lead-time spares, preparingfacilities, or acquiring technical data).

Specifically for human performance requirements, MPT planners might firstlevy maximum allocations of maintenance personnel, specific Air Forcespecialty (career) or skill distributions for those personnel, or maximumtraining resources available. As the system is developed, they might derivean optimum mix of these variables within these constraints.

In theory, human engineers function in the WSAP to convert theseconstraints into human engineering design criteria, and then to predict theMPT requirements consequences of the final design for the MPT planners. Thisfunction is, in essence, a cultural translation from the world of militaryinfrastructure to that of drafting boards or Computer Aided Design (CAD), andback. The operative question for this report is, "How to effect thistranslation?"

Ref. Chapter 2, Design For Maintainability; What Military Standards Do &Don't Sa, tools do not exist to enable this translation between engineeringdesign and MPT planning. Without a credible connection between design and MPTrequirements, levying MPT constraints and predicting MPT requirements are onlyritualistic exercises until the operational test and evaluation of aprototype. By then, significant design changes have large cost and schedulerepercussions; thus, they and so become infeasible.

1-I

In reality, MPT planning for new weapon systems is reactive to, ratherthan participatory with, design, and human engineering is without feedbackconcerning the human resources required to achieve satisfactory performance.

The fragmentation of human performance planning in the WSAP extends evenfurther. At least six different Air Force (AF) communities must particpate:

- Human Factors Engineering. Human factors engineering is usuallyoriented toward operability rather than maintainability.Analysts are typically matrixed to System Program Offices (SPOs)from engineering staffs.

- Reliability and Maintainability (R&M). R&M is usually orientedtoward reliability, rather than maintainability. Analysts areusually assigned under the Deputy Program Manager for Logistics(DPML) for each SPO.

- System Safety (S). S is usually oriented toward equipmentfailure and the designation of "human error" as a conclusivecause of safety incidents rather than a consequence of poordesign. Analysts are generally assigned from safety rather thanengineering staffs.

- Manpower (M). M is mostly performed by the operating MAjorCOMmand's [MAJCOM's] manpower plans office. M is augmented byinput from the Air Force Logistics Command (AFLC) manpower plansoffice for depot manning, from Air Training Command (ATC)manpower and technical-training personnel for training manning,and from the SPO for SPO manning. MAJCOM manpower plannerstypically do not have a complete picture of new technologies tobe fielded on the emerging system; nor do they have a knowledgeof human factors which might reduce manpower requirements.

- Personnel (P). P is managed by the AF Military Personnel Center(AFMPC), with coordination with the functional manager of eachAir Force Specialty [AFS]; the MAJCOM manpower, personnel,training, logistics, and operations personnel; and the ATCtechnical and operator training offices of primaryresponsibility. Identifying the most correct AFS to staff thenew system necessitates good human factors input about theskills, knowledges, and physical qualities of personnel which thenew system will likely require. No formal mechanisms exist forthis comunication.

- Training (T). Formal maintenance training is conducted by ATCat technical training centers or field training detachments.Training that cannot be performed by ATC is assigned toOn-The-Job Training (OJT) or specialized MAJCOM training. The3306th Training Development Squadron assigns maintainers to testand evaluation aircraft to become familiar with maintenanceactions and to conduct early phases of Instructional SystemDevelopment. Often they identify maintainability problems and

1-2

some are placed in service reports for correction.Unfortunately, redesign costs are often prohibitive this late inthe acquisition phase.

Each community traditionally uses separate analysis tools and data bases.(Personnel and training use few, if any, tools.) Data bases supporting eachdomain are fragmented and cannot easily be translated into the data bases ofanother domain. For example, maintenance manpower uses the Maintenance DataCollection System, personnel and training use the occupational survey andAdvanced Personnel Data Systems, safety uses lessons learned and MISHAP data

bases, and human factors uses data bases like the Center for AnthropometricResearch Data and predecessor SPO studies. Information gained in one domainis typically not passed along to others in any standard way.

The Armstrong Laboratory similarly fragments its development of newtools to serve its fragmented customer base. Different branches work indifferent disciplines. It is therefore not surprising that intradisciplinaryanalysis and data base/tool development is better understood an-funded thaninterdisciplinary. The result is that the communities have no standardmethods of communicating design-related data across domains.

In the authors' experience, this fragmentation has hurt all disciplines:

- Human Engineering has consistently suffered a lack ofprioritization within acquisition because of its inability toexpress quantitatively the operational and life-cycle costimplications of its recommendations.

- System Safety is now being tasked to address why human erroroccurs in accident investigations. It lacks appropriatetechniques.

- R&M is based on statistical predictions of failure and repairtimes which ignore the causal factors of human performance.Hence, it traditionally fails to address design formaintainability.

- Manpower estimates traditionally assume that new weapon systemmanning requirements will remain similar to the old requirements.Congress is now requesting guarantees of reductions of manpowerrequirements. Manpower planners lack techniques to predictrequirements for significantly new designs with non-traditionalmanning.

- Personnel, responding to keep systems near 100 percent manning atthe same time the overall force is reduced, wishes to utilizepersonnel across weapon systems in as few career fields aspossible. This will ensure that individual career ladders willnot be too small to manage effectively. Personnel also wantAFS aptitude and technical specialization requirements to bereduced so that demographic trends and force reductions can beaccommodated with the people likely to enter the military in

1-3

fewer career fields. Unfortunately, personnel has no tools todetermine what aptitude and technical specialization is requiredfor proposed new weapon system designs.

Training support and equipment are frequently acquired late dueto cuts in training development budgets in favor of other programpriorities. Training development might attain higher priority ifits relationship to human performance, and hence missionperformance, could be more explicitly demonstrated. In addition,training planning and development could be more efficient ifcareer fields could be more finely tuned based on common skilland knowledge requirements. However, there is no standard way todetermine skill and knowledge requirements from proposed newweapon system designs and communicate them to trainers.

The new Department of Defense Directive (DoDD), 5000.2, has recognizedthe need for the integration of human performance planning during the WSAP byincluding a separate section (equal with ILS) on Human-System Integration(HSI). HSI includes human engineering design, safety, manpower, personnel,and training. While this directive provides requisite policy to integratehuman-performance-related disciplines, it recommends no tools to do so.

CAMT was conceived to be a pragmatic, credible, analysis procedure usingan intrinsic data base to provide a cultural translation between engineeringand logistics within HSI. As an effective interdisciplinary tool, it wouldinevitably contribute to intradisciplinary analyses, but it is not intended tosupercede already existing tools and data bases. Rather, by more directlyconnecting subsystem design with Manpower, Personnel, Training, and Safety(MPT&S) consequences, we hope that individual MPT&S analysis tools willreceive better source data, and human engineering evaluations will moredirectly reflect life-cycle resource requirements.

Associated Concepts

Comparative Anatomy

In biology, a comparative anatomy analyzes species into common body parts1) to classify similar species into logically common ancestry, and 2) tocompare/contrast the functioning of each part in its adaptation to variousecological niches.

In maintenance, a comparative anatomy would analyze tasks into commonhuman performance parts 1) to classify similar tasks into logically commontraining sets and jobs, and 2) to compare/contrast the performance of eachtask part in its adaptation to the various weapon system designs.

Specifically, CAMT would:

- provide instant access to relevant maintainability design

1- 4

standards and lessons learned, as soon as the design is

conceived;

- aid in predicting MPT&S requirements for specific design options;

- provide a more logical approach to maintenance training curriculathan task-by-task teaching, i.e., teach common components to manytasks; and

- provide logical groupings of maintenance tasks into credible jobsand career fields as design becomes solidified.

The Scenario

The capabilities of CAMT are best understood through a realistic example.The following is a formal scenario that CAMT would specifically address.

A human engineering analyst on a design team is tasked to analyze one ormore proposed subsystem designs and predict the maintenance MPT&S consequencesof each. The analyst is specifically tasked to provide immediate designcritiques from a user-interface perspective, estimate the consequences of eachproposed design on MPT&S requirements, and then derive, in turn, the life-cycle cost implications of each.

The analyst has access to detailed design information and specificationsfor all subsystem components and their assemblies. The analyst also has alist of all required maintenance tasks as specified by the designer or the R&Mengineer. Finally, the analyst has the CAMT data base at fingertip access andis familiar with its contents.

Note that this tasking is not for traditional MPT analysis; that is, totrade personnel skill requirements, training times, job/career fielddefinition, and the number of people required in order to meet an establisheddesign's human performance requirements. Neither is this tasking fortraditional human engineering analysis; that is, to compare a proposed designwith human engineering specifications and guidance, then stop. Rather, it isto evaluate a proposed design using predicted "typical" MPT&S consequences.The emphasis is on design influence, not the derivation of the final MPTconfiguration. The tasking is thus to do HSI analysis, connecting thedisciplines.

Traditional Discipline Boundaries

Successful integration of segregated dsciplines requires an understandingof their traditional domains.

Human Engineering is the process of designing weapon systems, orof analyzing their designs, to be operator- andmaintainer-friendly. It minimizes the difficulty of requiredhuman performance, in turn minimizing learning and performancetime, propensity for human error, and the effects of stress fromfatigue, battle, temperature, motion, altitude, noise, orcollateral workload.

1 - 5

- Maintainability, applied to a weapon system, subsystem, orcomponent, is the probability that it will be retained orrestored to a given operational condition within a specified timeperiod with minimum resources (people, equipment, budget).Designing for maintainability means designing for the shortestpossible maintenance times and the largest probability of gettingit right the first time. Some aspects of maintainability designdo not directly relate to the maintainer: automated diagnostics,discard/replace options, functional location of test points, andsourcing of spare parts. However, most aspects do:accessibility, visibility, fail-safe orientation of replacementparts, minimum tool and crew requirements, potential safetyhazards, etc. These aspects are part of human engineering.

- System Safety is the prediction and minimalization of theconsequences of weapon system, subsystem, component, and userfailures. The subset of system safety which deals withconsequences of human failure is a part of human engineering.

Although human engineering, maintainability, and system safety all haveinfrastructure, life-cycle cost, and operational capability consequences,their primary influence is during weapon system design.

The AF Personnel function manages the acquisition,classification, development, and nurturing of people to fillauthorized positions AF-wide. The AF/Director of Personnelrecruits, classifies by skill, assigns, tests, promotes, andotherwise manages career progression. It strives to ensure AFmaintainers will be available in appropriate numbers, skills, andgrades to fill present and future manpower authorizations. Whentasks are reassigned between career fields (AFSs) or new ones areadded because of the new weapon system, it coordinates thechanges with the MAJCOM planning and logistics communities(MAJCOM/XPM and LG); the Air Staff functional managers; the AirTraining Command Technical Training staff (ATC/TT) and technicaltraining centers; and the AFMPC classification, assignment, andretention offices.

AFMPC tries to resolve AFS specialty restructures to meet theneeds of the specific weapon system while fitting in with the AFSstructure of other weapon systems. The career field restructuresmust consider the long-term capability to recruit, train, assign,and retain career field members satisfactorily. Therestructuring process could be faciliated by standard methods togroup tasks into jobs, and jobs into career fields based onrelated skill, knowledge, and task performance requirements.

The personnel function is also responsible for building thepipeline which takes people through recruiting orreclassification from other AFSs through initial orcross-training to assignment on the new weapon system. Personnelprograms the pipeline based on the authorizations by AFS that theMAJCOM manpower office programs into the Unit Manning Document.

1-6

- Maintenance Manpower determines manpower requirements necessaryfor a specific weapon system to meet the prescribed sortie rateand mission type. For most aircraft maintenance positions,Manpower uses Logistics COmposite Model (LCOM) studies. Thesestudies simulate the interaction of maintenance task assignmentto AFS with predicted system performance paramenters and proposedorganizational structure. Simulating the highest peacetime/wartime demand identifies the authorization levels at whichmanpower becomes the limiting factor in mission capability.These requirements, however, must be tempered by the end strengthallowed by Congress. Authorizations are placed into the CommandManpower Data System, which produces the Unit Manning Document.In turn, this document provides the goal of AFMPCreclassification and assignment actions. The more accurately theemerging weapon system is depicted and the AFSs classified, thebetter the manpower predictions will be.

- Maintenance Training raises the skill and comprehension levels ofavailable personnel to the proficiencies required by the manpowerpositions they are intended to fill. Technical training centersnormally train general skill and comprehension requirements.Field training detachments then train weapon-system-specificskills and knowledge. Airmen continue training via OJT andcorrespondence courses. As their skills and knowledge increasethey are upgraded progressively through "3-, 5-, and 7-skilllevels". These correspond to apprentice, specialist, andjourneyman levels of civilian job skills.

During acquisition, training planning determines the best methodof instruction for each anticipated task, then overseesprocurement of the required materials, equipment, facilities, andsupport personnel. ATC handles formal training and the usingMAJCOM handles OJT. Decisions about where and when to best teacha task could be improved if the specialties comprised relatedtasks, and the associated skill and comprehension requirementsfor each task were fully described.

MPT analyses constitute a more aggregate level of analysis than dosafety, systems, or human engineering analyses. That is, whilemaintainability and potential safety hazards can be directly associated withthe design of subsystems and components, MPT requirements are each theconsequence of many subsystem designs. Thus, while small engineering changesin one subsystem often make significant changes in overall systemmaintainability (subsystem design effects on subsystem maintainability areeasy to visualize and subsystem-to-weapon-system maintainability is a standardcomputation), their consequences for system MPT requirements are considerablyless direct and intuitive to predict.

Additionally, a significant driver of MPT requirements is inertia in MPTpractice. It may be too difficult to adjust the large, already existing humaninfrastructure of career fields, training courses, promotion tests, andmaintainers in the pipeline to take full advantage of

1-7

design-for-maintainability innovations. For example, although repairs wouldbe quicker, overspecialized maintainers would be left without work. Oralthough fewer skills would be required, the training courses would still bethe same length.

Consequently, traditional MPT planning has not focused on the suspectedlytenuous effects of detailed weapon system design. Now, however, Congressdemands firm MPT requirements predictions for proposed new designs.Infrastructure inertia--using approximately the same number and types ofmaintainers for the new system as were used for the old--will no longer betolerated. The design/MPT requirements link must be better understood.

CANT is attempting to bridge some of the larger gaps between thesehuman-related disciplines. "Comparing apples and oranges" will likely bedifficult--and the underlying concepts just as difficult to understand. Thegreater risk, however, is associated with greater potential payoff.

An important final note about confusing terminology: by traditional AFusage (Air Force Regulation 800-15), "human factors engineering" refers to allof these disciplines (design for operability, design for maintainability,design for safety, biomedical engineering, manpower, personnel, and training);"human engineering," refers only to the design disciplines. The new DoDD5000.2, however, uses "human factors engineering" to refer to what the AFtraditionally called human engineering, and "human-system integration" tomeanthe traditional AF human factors engineering. Because of this ambiguity, thisreport henceforth will no longer use the term, "human factors engineering."

Hierarchy of Design Influence

To aid visualization of the interrelationships between disciplines, noteFigure 1, the System Engineering Hierarchy of Design Influence. Thereferenced disciplines are ILS elements and subelements. Four levels ofdesign influence are shown, from subsystem components at the micro level, totheater-wide warfighting capability and life-cycle costs at the macro level.The bottom two levels are actually the borders of a continum of designinfluence which comprise traditional system engineering. Influence movesupward through the hierarchy, while mission requirements and constraints flowdownward. Traversing levels of the heirarchy usually requires a translationof some variables, in turn requiring assumptions about the values of others.

To focus on the HSI disciplines, begin by noting that MPT requirementsreside at Level Three, Infrastructure, along with the other ILS planningelements (e.g. support equipment, facilities, maintenance structure). Theymust derive solutions within the context of already existing administrativeand support structures (e.g., AFSs, technical training centers, location ofdepots, support equipment already in the inventory, levels of maintenance).They often deal with multiple weapons (e.g., a squadron of aircraft) at manylocations.

A weapon system's integrated logistics supportability directly influencesits mission performance (via deployability, availability, and sustainability)and life-cycle costs at Level Four. A weapon system's integrated logistics

1 - 8

mm. - .

a C)

= Z0 + 4-* 4*

~ CC>Zn

~ ~as4) ~.u

'~E~ 0~'u .

cn T~aE ~0

0-.4 1.4 C

supportability is, in turn, directly influenced by the designed-incapabilities of individual weapons at Level Two, System Design Specification(e.g., range, payload, information throughput, survivability, reliability,maintainability, immunity to psychological stress and fatigue, environmentalimpact, etc.).

Individual weapon capabilities are, in turn, determined by (1) thedesigned-in capabilities of their individual components (e.g. reliability,maintainability, corrosion control, user interface, parts standardization,etc.) at the bottom of the hierarchy, and (2) the integration betweenindividual components, assemblies, subsystems, and the weapon between LevelsOne and Two. During system engineering, trade-off analyses act horizontallyin the hierarchy. They compare various apportionments of design requirementswithin a single discipline between components (assemblies, subsystems) andvarious specifications of single components between disciplines.

Human engineering design for operability, maintainability, and safetybegins with weapon system components at the bottom of this hierarchy. Theirconsequences must undergo two major transformations to predict their MPTrequirements on Level Three, and three to predict life-cycle costs at LevelFour.

When surveying available acquisition tools, one finds a clustering at thetop and bottom of the hierarchy. The engineering community excels atallocating design requirements from weapon to component and integratingperformance capabilities from component to weapon. The logistics communityexcels at charting relationships between the functioning of each ILS elementand life-cycle cost (and, to a lesser extent, overall system performance).Transformations between Levels Two and Three, however, are mostly uncharted.

Human Performance Prediction

What is the common ground of human engineering, safety, manpower,personnel, and training? They all exist to ensure human performance. Humanoperators or maintainers must be able to accomplish a certain interaction withthe weapon system: first, to a certain level of accuracy; second, within acertain time period, and, third, with a controllable amount of collateraldamage if they err. Could human performance provide a common, possiblyquantitative, language to translate the effects of design between disciplinesand traverse the hierarchy?

Figure 2 depicts a narrower system engineering hierarchy of designinfluence specifically addressing human performance. Traditional humanengineering analyses proceed from the top downward; then study the effects ofdetailed system design on task steps. In contrast, logisticians study humanperformance on whole maintenance tasks to predict MPT requirements.Understanding design influence on the performance of whole tasks requires anunderstanding of design influence on the performance of each task step. Thereare approximately 10 to 40 task steps per task. Again, the transition betweenhuman engineering design effects on task step performance to the MPTrequirements of whole tasks is mostly unchartered. Why?

1 - 10

Human performance is innately more fallible, variable betweenindividuals, and changeable over time than other aspects of systemperformance. Therefore, it cannot be predicted by discrete mathematicalformulae, as can reliability or corrosion control.

However, it can be estimated probabilistically based upon emprical data.To predict human performance on maintenance tasks for a proposed new design,one must first collect human performance data from similarly skilled peopleperforming similar tasks (or portions of tasks) on similar designs undersimilar conditions. One may then compare each new design and its intendedenvironment with this "baseline" to subjectively adjust the human performanceinformation for design influence. One could similarly measure the MPT&Srequirements for attaining this human performance on the baseline, thenestimate the MPT&S consequences of each proposed design. This baselineinformation could be derived from (1) human engineering literature, (2) thefield, or (3) parametric research studies.

(System & Suprasystem

Functional Allocation

Q E Major Subsystem

Functional Decomposition

(Subsystem & Assembly

Task Analysis

EComponents

Figure 2. Human Performance Hierarchy

One could never remove all subjectivity from human performance estimationfor as yet unrealized weapon systems; hence, automatic analysis programs couldnever replace the expert judgement of a senior design analyst. However, toimprove the accuracy (and credibility) of human performance prediction, onecould support the expert with comparative data that allow straightforward,well-scoped judgements. How well are such data currently being supplied?

1 - 11

Current Manpower/Personnel/Training Practice and Its Limitations

Current ILS analysis practice is to use field data to define a "baselinecomparison system" (BCS)--a single composite of components from severalsimilar, currently fielded systems. Subject-matter experts look at the newdesign, choose which components of older systems are most like those of thenew one, collect relevant field data concerning logistics requirements on thebaseline components, then extrapolate how the change in design will changethese requirements in the new system. Maintenance task times, required skilllevels, AFS assignments, and overall training costs are established this way,then input to other MPT analyses.

Under this approach the effects of design change on MPT requirements aredifficult to estimate.

- The BCS data, once collected, is highly specific to the waponsystem under study; more will have to be collected for each newweapon system--a nontrivial task.

- The BCS data cover only one example of an alternative design foreach subsystem or component; they do not reveal how differingdesigns yield differing requirements.

- Since human performance data are only collected from alreadydeployed, similar systems, the greater the change in technologybetween the new and baseline systems, the less relevant thebaseline data will be. Perhaps more relevant baseline data wouldcome from a similar maintenance task on a very differentsubsystem, but one which used the new technology. (For example,would maintenance on a digital radio be more like maintenance onan analog radio or more like that on a digital radar signalprocessor?)

- The baseline data exist only at the whole-task level (e.g.,remove and replace, troubleshoot, calibrate) while design effectsare evident only at the task-step level (align coupling, applylubricant to gasket, attach probe). Collecting sufficient datato construct a baseline comparison system at the task-step levelwould be unwieldy. Therefore, each extrapolation of humanperformance from the baseline to the proposed system must includedifficult-to-audit, difficult-to-verify assumptions abou the

many design variables at the task-step level and theirinteractions.

Thus, the baseline-comparison-system approach, as currently practiced,requires large, complex extrapolations about subsystem design influence onhuman performance requirements from minimal baseline data. Could these largeextrapolations, covering many implicit design variables, be decomp d intomore muerous, smaller extrapolations, each with a more complete set ofsupporting data? If so, each extrapolation would be more straightforward andindividual errors would have less effect on the overall analysis results.

1 - 12

The CAMT Concept

The CAMT approach proposes that human performance could be betterestimated by using a level of analysis below that of whole tasks (which aretoo broadly defined to straightforwardly link to detailed design) and abovethat of task steps (which are too numerous and nuts-and-bolts oriented tostraightforwardly link to skills or manpower slots). CAMT assumes one could,instead, define common segments of tasks, "task primitives," that wouldrelate to both MPT requirements and human engineering design. The taskprimitives would serve as the foundation for a data base of typical MPTrequirements, covering several representative subsystem designs for eachprimitive.

Such a data base would have three obvious advantages over currentbaseline comparison systems. First, subsystem and component design influenceon MPT requirements would be easier to extrapolate for task segments than forwhole tasks. Second, the commonality of the primitives to many tasks wouldallow a single data base to be relevant to many weapon systems or subsystems.Third, the commonality of primitives to many subsystems would allow severalsamples of design-versus-MPT requirements for each primitive in the data base.Multiple samples provide the analyst with the opportunity to note patterns indesign consequences; these are useful for extrapolation to a new design.

The concept of an intermediate segment of human performance between atask and a task step is easy to postulate. Defining primitives for specifictasks in such a way to reap the above benefits, however, requires a morerigorous understanding of task primitive properties. A significant part ofthe CAMT concept development has thus been the evolution of a useful taskprimitive description.

The CAMT Task Primitive

Task primitives may be characterized as follows, progressing from generalto specific attributes:

- A task primitive is a set of desired human behaviors that couldlogically be taught together.

- A task primitive is expressed as one or more action verbs plusone or more behavior object classes, with optional modifiers. (A"behavior object" is any tool, article, equipment component, orsupplies with which the maintainer must come in contact whileperforming the task.)

- A task primitive is not a task step, although it conceivablycould be identified from a single step. Most task primitives arebased on several, usually non-conseciitivp steps which may bescattered throughout the task procedure. Nor is a task primitivea whole maintenance task--unless the task is very simple and itsbehavior objects can be generalized. Several task primitivescomprise each maintenance task, and many maintenance tasksinclude the same task primitive.

1 - 13

(Examples of task primitives are (a) remove and install safety wire; (b)remove and replace oil drain plug; and (c) set and operate jack toraise/lower vehicle.)

- Each task imposes on the task primitive certain Design-SpecificCharacteristics and Constraints (DSC&Cs) for human performance.These are atypical human performance and resource requirementswhich result specifically from system design.

(Examples of DSC&Cs are (a) safety-wire accessibility; (b) the type offrame a jack must support (X, perimeter, unibody); or (c) drain-pluglocation.)

- Associated with each task primitive is a set of basic skills,practiced skills, and comprehension requirements.

-- Basic skills are commonly required human performancecapabilities, often learned prior to active duty.

(Examples include (a) using a screwdriver; (b) lifting heavy objects; or(c) reading technical orders. Attainment of basic skills may bereflected by achieving a minimum Armed Service vocational AptitudeBattery (ASVAB) score.)

-- Practiced skills are the unique proficiency requirements ofeach task primitive and must be trained.

(Examples include (a) tightening the oil drain plug to within a narrowrange of torque values; or (b) ensuring that the jack contacts 4 balancedpoints on the vehicle frame.)

-- Comprehension requirements involve theoretical understanding.They may either be common to many task primitives (such asunderstanding the basic principles of jet engine operation)or unique to one task primitive (such as understanding theprinciples of safety wiring).

- A useful set of task primitives must meet three criteria withintheir maintenance domain.

-- Generality. Each task primitive contributes to many diversetasks. Therefore, data indexed by it is useful in manycontexts.

Orthogonality. No two task primitives are highly correlated;there is never confusion about which of several taskprimitives apply when decomposing a task. This implies thereis no ambiguity about action verbs or behavior objects.

Completeness. Each maintenance task can be completelydecomposed into task primitives with no task segments leftover.

1 - 14

The initial identification of task primitives requires task analysismethodology. It entails a thorough examination of a task's steps, behaviorobjects, DSC&Cs for human performance, possible errors, and skill andcomprehension requirements. However, as more tasks are examined within themaintenance domain, fewer new task primitives should be needed. Fairly soon,a newly examined task may consist entirely of previously identified taskprimitives. Only the DSC&Cs would be new.

When such a task is found, it means that training provided on thepreviously identified task primitives, if general enough, would be sufficientto establish proficiency on this task also. (This assumes DSC&Cswould be presented via job aids.) It also means that the MPT&S requirementsof the previous tasks could be extrapolated to the new task, primitive byprimitive. Finally, it means that the number of common primitives between twotasks could serve as a metric of task similarity to support AFS structuring;i.e., AFSs could be formed from task groupings that contain many common taskprimitives.

The CAMT Data Base

The purpose of the CAM Tmethodology is not to define another maintenancetask taxonomy, but to serve as the foundation for a powerful, accessible database to support concurrent engineering and integrated logistics supportplanning. As with current baseline comparison systems, CAMT data would becollected from field or laboratory experience with already existing designs;however, it would be collected for common task segments-rather than wholetasks--and sampled across several diverse designs--rather than just one. Weassume that training on task primitives would also occur across severaldesigns, rather than just a single weapon system, and would collecttraining-relevant data appropriately.

Data dictionaries would be built around 1) task primitives and 2) basicskill and comprehension requirements (which could apply to many taskprimitives). While the amount of data associated with each entry is large,it is readily collectable.

Associated with each basic skill and comprehension requirement would be:

- the notation of whether it could and should be trained (versusacquired through recruitment);

- the typical time to train, if appropriate;

- the typical method of training and associated costs; and

- the typical retention time when not used.

Associated with each task primitive would be:

- its definition;

- its associated basic skill, practiced skill, and comprehensionrequirements;

1 - 15

- general proficiency requirements (usually required accuracies)for its practiced skills;

- its potential human errors and their consequences over diversedesign configurations;

- a sample of diverse relevant design configurations, along withthe following information for each:

-- its DSC&Cs,-- its typical time required for task primitive performance, and-- any design-specific proficiency requirements;

- especially relevant lessons learned and design guidelines, 1)indexed from human engineering standards and the AFlessons-learned data bank, and 2) derived from the DSC&Cs ofother systems and their known human performance consequences; and

- the traihing required to achieve proficiency on its practicedskill requirements for several diverse relevant designconfigurations. Specifically:

-- The typical time to train- The typical method and cost of training

The typical retention time for proficiency when not used.

The "typical method and cost of training" refers to the resourcesrequired to bring an entire class of students to proficiency, assuming theyall meet minimum requirements. The "typical time to perform" refers to thetime required by the slowest proficient (not marginal) maintainer. The"typical retention time" refers to the length of time proficiency requirementscould be met without refresher training, when the task is not practiced.

Assuming task primitives could be culled from maintenance tasks and thatthe above data base could be populated, an analysis comparing new designs witholder ones in terms of task primitives, rather than tasks, would offer manyadvantages.

- Since task primitives are common to many systems, a comparisonsystem based on them would require only the CAMT data base. Theattributes of any task within the maintenance domain could beassembled and extrapolated from those of its segments-the taskprimitives--for any design. This would eliminate the requirementto collect a new set of idiosynchratic baseline comparsion datafor each new weapon system, saving analysis time and resources.

- Although there are essentially an infinite number of subsystemsand components to be maintained (as in, "remove and replace boxX," "test and verify card Y"), there are a finite number ofbehavior object classes (such as tubes, cables, couplings, cards,pumps, fasteners, fastener tools, measuring devices, inspectiondevices, cleaning agents, etc.). A data base for all task

1 - 16

primitives within a maintenance domain, while large, should beconsiderably smaller than a data base covering entire maintenancetasks at the subsystem level (such as those maintained by the AFOccupational Measurement Squadron).

Since task primitives would be common to many maintenance tasks,and therefore many designs, the task primitive data basecould easily include multiple baselines. Human performance couldbe measured on at least three diverse designs; details of thedesigns themselves would be included. Admittedly, it would behighly unlikely for a design featuring substantially newertechnology to resemble any of the design baselines.Nevertheless, the human engineering analyst could readily accessexamples of how differences in design correlate with differencesin human performance.

Since task primitives, by definition, are more specific thantasks, there would be fewer total influencing variables on humanperformance at the task primitive level. The effects ofalternative design approaches on task primitive performance, asopposed to task performance, should be more direct and intuitiveto estimate. (In corollary, data down to the level of individualtask steps would be closely tied to component physicalconfiguration, but would be intuitively distant from basicskills, training times, skill retention, manpower slots, etc.)

Current human engineering design guidelines are contained ingeneral standards. Theoretically, each design feature of eachcomponent should be checked with hundreds of pages of guidelines.This process is too tedius during iterative design andconsequently doesn't occur until critical design review-if atall. The Air Force Lesson-Learned Data Base has a similarproblem. Its index for maintainability lessons runs over 50pages; no designer is likely to read all the lessons beforetackling a project. Lessons-learned and design guidelines,however, could be specifically indexed to task primitives viatheir action verbs and behavior objects, such that as soon as ananalyst calls out an primitive, s/he has ready access to humanengineering criteria specific to the design in question.

If task primitives are indeed defined to be general, orthogonal,and complete, they offer an efficient approach for organizingtraining. Technical schools could teach all the task primitivesand underlying skills which comprise an AFS's maintenance tasks.Teaching them on multiple designs representative of the realworld (e.g. on part task trainers) would produce versatilemaintainers who would need only a well-written job guide to adaptto a new subsystem. The generality of the task primitive, asopposed to the equipment specificity of whole maintenance tasks,makes it an ideal foundation for such an approach.

1 - 17

Structuring the CANT Data Base: An Example

Perhaps the best way to convey the concept of "task primitive" is byexample. The discussion that follows describes methods for structuring a TaskPrimitive Dictionary and a Basic Skill and Comprehension Requirements Diction-ary. The data are gathered through observation of task performance andthrough interviews with technicians thoroughly familiar with the subjecttasks. Because of the similarity of task primitives across tasks within asystem, the process gets easier as more tasks are analyzed. Once dictionariesfor one system have been compiled, CANT analysis for a similar system is mucheasier.

Removal and Replacement of the F-16 Jet Fuel Starter

The Jet Fuel Starter (JFS) is connected to the Accessory Drive Gearbox(ADG) by means of a V-band coupling (clamp). Before the coupling is removed,several tube nuts and electrical connectors must be disconnected. After thecoupling is removed, the JFS is pulled back from the ADG and slowly lowered tothe ground.

To begin removal, the technician first disconnects three tube nuts andremoves the drain tube connected to the bottom of the JFS. One of these threetube nuts is connected to the purge tube, which runs between the drain tubeand the top of the JFS. Next, the technician disconnects the upper tube nutof the purge tube. The remaining disconnections pertain to: the electricalconnector (cannon plug), ignition cable connector, air tube nut, main fueltube nut, and start fuel tube nut. (The air tube nut and the ignition cableconnector have safety wire that must be removed.) When these disconnectionshave been accomplished, the JFS is free of the aircraft. All that remains isto loosen the V-band coupling and remove the unit.

To reinstall the JFS, the procedure is followed in reverse order. AnyO-rings, seals, or gaskets that are damaged or deteriorated are replacedbefore JFS reinstallation. The tube nuts, ignition cable connector, andV-band coupling nut are tightened to specified torque values. The air tubenut and the ignition cable connector are safety wired.

To determine the task primitives for this task, the analyst will firstlist the typs of behavior objects that the technician manipulates whileperforming the task, namely:

1. Compression fittings (tube nuts)

2. Cannon plug

3. Safety wire

4. V-band coupling

5. O-rings, seals, and gaskets

6. Torque wrenches

1 - 18

7. Crow's feet and extensions

8. Safety-wire plier-s

9. JFS

10. Cannon-plug pliers

11. Open-end wrenches

Next, the analyst will consider combining some of the types of behaviorobjects to develop topics that would be taught together in training for thistask. For example, one would not consider teaching about compression fittings(1) without also teaching the use of crow's feet (7) and open-end wrenches(11). Therefore, one should consider that numbers 7 and 11 are subsumed bynumber 1. One might also consider teaching the use of torque wrenches whenteaching about compression fittings. However, torque wrenches may be usedwith other types of connectors and fasteners, and there is enough subjectmatter to be taught about the use of torque wrenches that a separate instruc-tional segment on this topic would be justified.

Using similar reasoning, 10 should be combined with 2, and 8 should becombined with 3. Number 9 (JFS) is not a type of behavior object-it is toospecific to the task. The analyst should consider: "to what class does itbelong?" The JFS is a very heavy object. The job guide specifies that twopersons should be used to lower it to the ground and raise it into position.Although one very fit and strong individual could conceivably handle it, thiswould not be a safe practice. There are many hints that can be conveyed abouthow heavy objects should be handled. Therefore, number 9should be changed to"manipulate heavy objects." The final list of task primitives for "remove andinstall the F-16 jet fuel starter" then becomes:

1. Disconnect and connect compression fittings.

2. Disconnect and connect cannon plugs.

3. Remove and install safety wire.

4. Remove and install V-band couplings.

5. Remove and install 0-rings, seals, and gaskets.

6. Manipulate heavy objects.

7. Use torque wrench.

A technician who is competent in these task primitives is adequatelyprepared to remove and install the JFS, with the aid of job guides. Many ofthese same task primitives are present in other maintenance tasks. Forexample, the list of task primitives would be similar for the F-16 task:"Remove and install the engine-driven hydraulic pump."

1 - 19

Errors and Error Consequences

The next step is to list the errors associated with each task primitiveand the consequences of those errors. The errors listed at this point are notsystem-specific. They are errors that would be expected if this same taskprimitive were observed in a different system. An example, for the fir-t taskprimitive of the JFS task, is as follows.

1. Disconnect and connect compression fittings.

1.1 Use of the wrong tools or overtorquing could result in rounding

the nut.

1.2 Incorrect use of crow's-foot extensions could make connectionand disconnection difficult or impossible.

Design-Specific Characteristics and Constraints

DSC&Cs are task primitive factors that are unique to the task--that setapart the task primitive within this task from the same task primitive when itappears in another task (or even another system). For task primitivesinvolving the disconnection of connectors or the removal of fasteners, suchinformation as how many are disconnected (removed), and mechanical and visual

access difficulties, should be recorded. For the first task primitive of the

JFS task, an example of task-specific constraints is as follows.

1. Disconnect and connect compression fittings.

1.1 Seven compression fittings need to be disconnected and con-nected.

1.2 The upper purge tube nut and the air tube nut have difficultvisual and mechanical access because of severely limited space

above the JFS.

1.3 Safety wiring of the air tube nut is accomplished in a severelylimited space.

Design-specific characteristics of a task primitive include such data asacceptable voltages at test points, accuracy and precision requirements (ingeneral terms), personnel and equipment safety hazards, and possible errorsand their consequences. Examples of design-specific characteristics for thefirst task primitive of the JFS example are as follows.

1. Disconnect and connect compression fittings.

1.1 All seven compression fittings must be torqued to specifiedvalues during installation. The torque values range between 75

and 370 inch-pounds.

1 - 20

1.2 Design-specific errors and their consequences are as follows:

1.2.1 Failure to disconnect one of the fittings during JFSremoval could result in damage to the tube. Bent tubeswill be hard to reattach. Kinked tubes will have to bereplaced.

1.2.2 Failure to distinguish between the main fuel tube nutand the start fuel tube nut could result in impropertorque.

1.2.3 Failure to distinguish between the drain tube nuts andpurge tube nuts could result in improper torque.

1.2.4 If the upper purge tube nut is permitted to slip downthe tube before the JFS is reinstalled, the nut must belocated and brought to the top of the purge tube beforeit can be reconnected. This operation is difficult andtime-consuming because the purge tube runs behind theJFS (where access is limited).

1.2.5 If any tubes are bent during JFS removal, the tube nutwill not fit squarely on its threads, and it will behard to start the nut.

1.2.6 The air tube runs from the inboard side of the JFS tothe outboard. When disconnecting or connecting the airtube, some technicians tend to turn the nut in the wrongdirection. Depending on how long it takes the tech-nician to realize the error, the step can be verytime-consuming.

1.2.7 Failure to use the two-wrench method when connecting ordisconnecting some of the tube nuts can damage the tubesto which they are connected.

The design-specific errors listed above are the types of errors thatexperienced technicians should notice. They are essential for safety analystsand writers of technical orders; however, they are too specific to be of muchuse to MPT&S analysts. The CAMT analyst should examine these design-specificerrors to determine whether some or all would be moved to the lists of errorsassociated with task primitives. The ;uestion to be answered is: CaW aprinciple be derived from one or a group of these errors that, if conveyedthrough training, can reduce the frequency of this type of error? Thedesign-specific errors listed above can all be translated into principles thatneed to be taught when presenting a course segment on the topic of compressionfittings. The principles are as follows.

1.2.1 Before removing any component from which you havedisconnected several compression fittings, review thejob guide to ensure all necessary fittings have beendisconnected.

1 - 21

1.2.2 Before torquing any compression fitting, review the jobguide (especially the illustrations) to ensure theappropriate fitting and torque values have beenidentified.

1.2.3 Same as 1.2.2.

1.2.4 Before installing a new component (or reinstalling anold one), ensure all tube nuts are at the ends of theirtubes.

1.2.5 When removing a component, avoid snagging (catching) atube on some portion of the component. If a tube isbent, the nut will not fit squarely on its threads whenreattaching the tube.

1.2.6 When the male union to which a compression fitting isattached projects away from the maintainer, there is atendency for maintainers to turn the nut in the wrongdirection when removing or connecting the tube nut.Identify all such cases at the outset, so time is notspent turning the nut in the wrong direction.

1.2.7 (This error is already stated in general terms. It canbe moved without modification to the general taskprimitive errors.)

The next step for the CAMT analyst is to restate the design-specificerrors in general terms and add these errors to those already on thetask-primitive error list. For the first two design-specific errors, therevised error statements should begin with the words, "Failure to review thejob guide ......

Skill and Comprehension Requirements

The skill and comprehension requirements are specific topics andbehaviors that, when mastered, prepare technicians to perform task primitives.However, during MPT&S analysis, decisions may be made to fulfill theserequirements through either selection, training, or performance aiding.Examples of skill and comprehension requirements for the JFS task are listedbelow.

1. Disconnect and connect compression fittings.

1.1 The two-wrench method; how and when to use it.

1.2 Principles of wrench use.

1.2.1 Choosing the type of wrench appropriate to the task(open-end, box, adjustable, crow's feet, tubingcrowsfoot).

1 - 22

1.2.2 Choosing the right size wrench.

1.2.3 How to use an adjustable wrench.

1.2.4 Keeping the plane of the wrench perpendicular to theaxis of the tube, nut, or bolt.

2. Disconnect and connect cannon plugs.

2.1 Recognition of the type of cannon plug and how to remove andattach it (screw, quick-disconnect, bayonet).

2.2 When and how to use cannon-plug pliers.

2.3 Types of keyways.

3. Remove and install safety wire.

3.1 Types and sizes of safety wire.

3.2 Safety-wiring methods (single-strand, double-strand).

3.3 Safety-wiring precautions.

4. Remove and install V-band coupling.

4.1 Types.

4.2 How they are constructed.

4.3 How they are loosened and tightened.

4.4 How to keep them out of the way when mating two components.

5. Remove and install O-rings, seals, and gaskets.

5.1 Their careful removal to preserve serviceability.

5.2 Their installation to provide a tight seal.

6. Manipulate heavy objects.

6.1 Principle of center of gravity.

6.2 Posture when lifting or lowering a heavy object.

6.3 Precautions to prevent injury.

7. Use torque wrench.

7.1 Types of torque wrenches.

1 - 23

7.2 How to set them.

7.3 Use of crowsfoot and extensions. The effect of extensions ontorque measurements.

Once this structure is set, the Task Primitive Dictionary and the BasicSkill and Comprehension Requirements Dictionaries could be populated with dataderived from interviews with technical school instructors and line-levelsupervisors of the responsible maintainers. Appropriate methodology will beaddressed in a future study.

Human-Systems Integration Analysis with CAMT

Assuming a CAMT data base has been built, how will it support thescenario presented on page 1-5? The human engineering analyst is tasked topredict the MPT&S consequences for each of several proposed designs of asubsystem. These predictions will be used primarily for design evaluation andsecondarily as inputs to more traditional MPT&S analyses. The analyst isgiven the CAD drawings of each design and the list of maintenance tasksassociated with each. The analyst has the CAMT data base at fingertip accessand should be familiar with it. "Subsystem" could refer to major subsystems,individual components, or any level of system design in between; "subsystemdesign" refers to how this level of system design is comprised of componentsfrom the next lower level.

How will the human engineering analyst generate a comparative anatomy ofmaintenance tasks, and what will s/he learn from it?

Referring to Figure 1, although a CAMT may traverse all four levels ofthe Hierarchy of Design Influence to some extent, its main value lies intraversing between Levels Two (System Design Specification) and Three(ILS Requirements). A CAMT will augment already existing tools whentraversing between Levels One and Two, and between Levels Three and Four.

Figure 3 outlines a general maintenance anatomy through which the humanengineering analyst will navigate when evaluating proposed subsystem designs.For each design option, the analyst will:

- break all required maintenance tasks into their constituent taskprimitives,

- look up the typical human performance and resource requirementsfor each constituent primitive,

- adjust these typical requirements to address the design underconsideration (taking into account all DSC&Cs),

- aggregate these human performance and resource requirements backto task level, and

1 - 24

- aggregate these requirements across all relevant maintenancetasks to the subsystem level.

The subsystem totals serve as comparison criteria for competing designoptions.

CAMT Subsystem Design Analysis. For Each Design Option,Aggregate Task Times, Errors, Design Guidelines, MPT Resources

Lcvel 2:SUYTESubsystemAnalysis

Level 1: T.P. #X T.P. #Y T.P. #Z

TaskAnalysis

Estimate CAMT •_•_•DSC&C Typical-DataEffects: Base w. 3 Rep.Designs/T.P.

Figure 3. Overview: CAMT Analysis

(Although the top half of Figure 3 is reminiscent of traditional taskanalyses using single comparison baselines, the bottom half refers toaccessing multiple baselines using the unique task primitive data base.)

The analyst begins with a subsystem design option to evaluate and a listof all maintenance tasks anticipated for this design. The specific stepsfollow in detail.

1. Maintenance Task Anatomy. Referring to the task primitivedictionary, decompose each task into its constituent primitives;then access their human performance and resource attributes in theCAMT data base. Via direct study of relevant CAD layouts, comparethe proposed new design with the various baseline designs and theirDSC&Cs in the task primitive dictionary. Identify any new DSC&Csfor human performance associated with the proposed design.

1 - 25

2. Basic Design Feedback. Via direct study of the relevant CADlayouts, evaluate the prnposed design using the design criteria andlessons-learned associated with each relevant task primitive in thedictionary. For any discrepancies, recommend specific designmodifications. (The task primitive dictionary information should bespecific enough to allow this.)

3. Task-level Safety Data Generation. For each task primitive, itemizethe potential errors and their consequences from the task primitivedictionary. Next, compare the relevant CAD layouts of the newdesign with the various baseline designs in the task primitivedictionary. Estimate any potential design-specific potential errorsfor the new design. Aggregate all potential errors to the tasklevel.

4. Task-level Maintainability Analysis. First, via consultation withthe designer, note any design-specific practiced-skill requirementsfor each task primitive (i.e., more/less stringent accuracyrequirements or unique design-driven performance criteria). Then,by comparing the relevant CAD layouts of the new design with thevarious baseline designs in the dictionary, estimate the typicaltime for each task primitive's performance within the new design.Extrapolate this from the typical times to perform listed for thebaseline designs, taking into consideration any DSC&Cs and anydesign-specific proficiency requirements noted above. Aggregate totask level.

5. Task-level Personnel Data Generation. Aggregate the basic abilityrequirements listed for each component task primitive to the tasklevel.

6. Task-level Manpower Data Generation. For each task primitive, bycompare the new design with the multiple baselines to estimatewhether more than one person will ever be simultaneously requiredfor successful performance. Note for redesign, if possible;otherwise aggregate such instances to the task level.

7. Task-level Training Data Generation. For each task primitive,(using the task primitive dictionary) itemize as training goals thecomprehension requirements, practiced-skill requirements (general ordesign-specific, if appropriate), and any basic-skill requirementsthat must be trained. For each of these, list typical trainingmethods, resources required, and times to train, as inputs totraining planning and life-cycle cost estimates. Subjectivelyadjust the training times, if appropriate, to compensate fordesign-specific proficiency requirements (e.g., training to higheraccuracy requires more practice). Aggregate to the task level,taking care to count each basic skill, practiced skill, andcomprehension requirement only once per task.

8. Subsystem-level Safety Data Generation. Aggregate potential errorsfrom all tasks to the subsystem level for use as input to systemsafety analyses.

1 - 26

9. Subsystem-level Maintainability Analysis. Sum typicalmaintainability times from serial tasks (e.g., remove-and-replace X,+ troubleshoot x, + verify X) to derive typical times to repair forvarious faults called out in the Failure Modes and Effects Analysis.Compare these with maintainability allocations to determine theadequacy of the design.

10. Subsystem-level Personnel Data Generation. Aggregate all basic-skill requirements to the subsystem level to serve as a total ofmaintainer personnel qualifications for this specific design.

11. Subsystem-level Manpower Planning Strategy. Propose clusters ofthese maintenance tasks to be handled by single AFSs, using thefollowing strategy. Aggregate task primitives across all subsystemtasks to form a subsystem listing. These can be compared with taskprimitive listings for other subsystems. Those subsystems whosetasks have the most task primitives in common may be assigned to asingle maintenance specialty. Or, if they must be assigned topreviously existing specialties, compare the task primitive listingfor this subsystem with the aggregate requirements of existingspecialties to choose a best match. Also, aggregate all occurenceswhere more than one person must be present to accomplish a taskprimitive.

Note: Instead of aggregating maintenance tasks to subsystem level,then grouping subsystems for assignment to a single specialty (basedon commonality of task primtives), one could remain at the tasklevel and cluster into specialties independent of subsystems. Forexample, the commonality of task primitves could conceivably lead usto the definition of more generic AFSs in cables, lubrication,electronic power supplies, signal processors, etc.

12. Proficiency Retention Analysis. Compare the expected time betweentask primitive performance with the expected retention time for taskprimitive performance. The expected time between performance is theinverse of the expected frequency of performance. The expectedfrequency of performance is, in turn, the sum of the expectedfrequencies of performance of all maintenance tasks which includethe primitive.

Hence, for each task primitive, itemize all maintenance tasksrequiring its performance in the specialty as just defined. Add thefrequencies of occurence for all these tasks, then compare(inversely) with the typical retention time for the task primitive'spracticed skills. If a maintainer would perform the task primitiveless frequently than required to maintain proficiency, it would bebetter to aggregate tasks differently for his/her specialty.Otherwise, refresher training must be provided, with its attendantincreased resource requirements.

1 - 27

13. Subsystem-level Training Data Generation. Aggregate thecomprehension requirements, practiced-skill requirements, basicskill requirements to be trained, accuracy requirements, typicalmethods of training, typical resources required, and typical timesto train to the subsystem level. Eliminate redundancies.

Note: It would be most efficient to structure training around taskprimitives rather than specific weapon systems; a new weapon systemwould thus only require new training on newly-defined taskprimitives or unusual DSC&Cs.

14. Job Aids. Once a design is finalized, collect all of its DSC&Cs,accuracy requirements, and potential errors, as well as thethree-dimensional design information itself, as the foundation forspecific on-the-job references, be they in the form of positionalhandbooks, technical orders, Integrated Maintenance InformationSystem programs, etc.

What has the analyst learned upon completing these analyses? First,s/he has derived the following information.

1. A listing of all human engineering design criteria andlessons-learned relevant to this specific subsystem design and anassessment of how well this specific subsystem design heeds them

2. A detailed listing of all potential human performance errorsassociated with this specific subsystem design

3. Estimated performance times for every maintenance task ormaintenance action (a group of sequential tasks) associated withthis specific subsystem design

4. An aggregation of all basic-skill and comprehension requirements forhuman performance associated with this specific subsystem design'smaintenance

5. An aggregation of all maintenance tasks where this subsystem designrequires more than one maintainer (specified to the task primitivelevel)

6. Proposed optimum clusters of maintenance tasks for assignment toideal AFSs, or proposed optimum clusters when forced to assign totraditional AFSs

7. A retention analysis of each proposed task cluster

8. A total of all practiced-skill, basic-skill, and comprehensionrequirements for human performance that must be trained for thisspecific subsystem design, and a reasonable estimate of the totalcost and time for this training

1 - 28

9. A recommended technical school training approach for maximu=training transferability: organize curricula around task primitivesand train each on multiple design configurations

10. DSC&Cs and practiced skill requirements

Returning to the original scenario, information items 2, 3, 4, 5, and 6can all be used as MPT&S-related design evaluation metrics: the less of eachitem, the better the design.

Additionally, this information provides a foundation for further HSIanalyses.

- Result 1 serves as traditional human engineering analyses, butprovides immediate, focused design feedback pertinent to eachtask primitive. It eliminates the tedious indiscriminantapplication of all human engineering standards andlessons-learned to all parts of the weapon system.

- Result 2 gives design-related input to traditional system-safetyanalyses.

- Result 3 gives design-related estimates of subsystemmaintainability for all associated tasks.

- Result 4 can be compared to the anticipated demographics ofrecruits during personnel analyses. If the requirements are toohigh, the subsystem must be redesigned.

- Result 5 documents some design-related high-drivers of totalmanpower requirements.

- Results 6 and 7 provide a good strategy for assigning tasks to AF

specialties in order to minimize training requirements.

- Results 3, 5, 6, and 7 provide design-related inputs to LCOM.

- Results 8 and 9 provide design-related input into InstructionalSystems Development.

- Result 10 plus the actual CAD displays of the proposed designprovide excellent content for job aids.

- Once CAMT analyses have been performed on subsystems from manyweapon systems, their results could aid maintainer careerplanning (personnel management). To ensure optimum use of pastexperience on new assignments, a maintainer could be programmedinto a series of assignments featuring most of the same taskprimitives but with substantially different DSC&Cs, or onesfeaturing a significant percentage of new task primitives eachtime while nevertheless keeping a majority of task primitivesbetween succeeding assignments the same.

1 - 29

CAMT Applicability

When would a CAMT analysis be performed in the weapon system acquisitionprocess? The purpose of the CAMT methodology is to integrate NPT&Sconsiderations directly into the weapon system design process. It would thusbe used throughout the system engineering process from the first appearance ofthree-dimensional configurations on a CAD system until actual prototypes arebuilt for more direct maintainability testing.

Figure 4 sumarizes what a CAMT system would and would not do.

CAMT is notCA i

1. A CAD tool like Crew Chief, nor 1. A tool to translate between levels ofa base-level simulation like LCOM design analysis, from design to MPT&S

2. An MPT&S tradeoff tool, reacting to 2. A design analysis tool, comparinga fixed design like Isoperformance "typical" MPT&S consequences of

proposed design options

3. A modeling shell, performing 3. A data base, structured for easy accesscalculations and logic on data to be and utility across many applications atinput by the user like SUMMA the user's workstation

4. Abilities based, focused primarily 4. Action based, focused on taskon maintainer characteristics like characteristicsFleishman

5. A taxonomy, dealing with overall 5. An anatomy, dealing with common tasktask traits for comparison pieces for comparison

Figure 4. Comparative Anatomy of Maintenance TasksFoculs

First Steps

Many feasibility issues arise. Our present ezfort had a very limitedscope and addressed only the most fundamental issue: given a set ofmaintenance tasks on diverse systems, could we define a well-scoped set oftask primitives which are general, orthogonal, and complete? We alsoperformed a literature review for precedents for CAMT methodology.

1 - 30

SECTION II. FIELD STUDY

The purpose of this study was to test the feasibility of defining taskprimitives for three maintenance tasks and to determine the extent to whichtask primitives defined within one weapon system may be found in the same orsimilar tasks within other weapon systems. Three tasks within three weaponsystems were examined. The tasks were removal and replacement of: the fuelpump, starter, and enginer turbine. The weapon systems were the J-57 engineof the KC-135, F-101 engine of the B-lB, and T-56 engine of the C-130.

Data Collection

A three-man team interviewed four Strategic Air Command (SAC) and twoMilitary Airlift Command (MAC) engine-shop crew supervisors. For each task,the crew supervisors took the team members to the shop floor, showed them theconnections and components involved, and described the task and any problemsencountered in its accomplishment.

Next, the group went to a conference room to discuss each step of thetask. If the data collection team had difficulty in visualizing any of thesteps, they returned to the shop floor to watch the crew supervisorsdemonstrate the steps.

For each step, the following data were obtained:

- Step description- Behavior objects (equipment, tools, and supplies with which

technicians interact during the task)- Constraints- Possible errors

As the data were obtained, they were written on a whiteboard or blackboard, sothat all participants could agree on the completeness and correctness of thedata.

After obtaining these data for the first remove-and-replace task (theB-lB fuel pump), a discussion was conducted concerning the major curriculumitems that would need to be covered in a basic course designed to preparetechnicians to work in the engine shop. The following topics were mentionedby the crew supervisors.

- B-nuts (compression fittings)

-- Installing unions (jam nuts, thread protrusion)-- Tube alignment (preventing cross-threading)-- The two-wrench method (when, where, and how to use)-- Torque wrench use (types, setting methods, calculating torque)-- Safety wiring

2-1

- Electrical connectors

- Types- Installation-- Torquing- Safety wiring

- Common hardware

-- Nuts and bolts-- Screws- Couplings and clamps-- Pins-- Seals

- Cleaning agents and lubricants

- Safe use of

- Powered and non-powered Auxiliary Ground Equipment (AGE)

- Inspection procedures

- How components look when they are unacceptably cracked, peeled,scored, etc.

- Common tools and measuring equipment

The crew supervisors were especially enthusiastic about using part-tasktrainers to teach such topics as how to remove, install, and torque nuts insituations where access is severely limited-and to provide practice in theseoperations. They also favored providing considerable part-task practice insafety wiring, extraction of panel screws, and removal and installation ofgaskets and seals. One of the crew supervisors strongly favored teachinginspection criteria. He said that when he first came to the shop, he had noidea of what an unacceptable equipment item looked like-how bad was too bad.

Another discussion on training topics was conducted after data weregathered concerning the removal and installation of the starter on the KC-135.The following topics were mentioned.

- Lubrication

-- Types of lubrication-- When not to use

--- Do not use oil where fuel is present--- Do not use fuel where oil is present-- Do not use fuel or oil for hydraulic or pneumatic

connections-- How to lube various types of components and connectors-- What happens if you fail to lube

2-2

-- Toxicity of lubricants

-- How to dispose of used petroleum products

- Torquing

-- How to set the wrench-- Consequences of over- or under-torquing-- When recalibration of the torque wrench is necessary-- False reading when the crow's foot contacts a component or

line before specified torque is reached

- Safety wiring

-- Neutral pulls-- Consequences of too much or not enough twist-- Consequences of wrong type of safety wire-- Do not cut pigtail on the diagonal

- Use of socket wrench

-- Getting the socket on straight (extensions)-- Need for deep-well sockets-- Knuckle protection

--- Use the heel of the hand--- Anticipate a sudden break-loose

-- May have to remove another component to gain access

After data were gathered for all the SAC tasks (B-lB and KC-135), thestudy team and crew supervisors constructed a list of taskprimitives designed to cover all the tasks performed in the engine shop. Atthis point, all three tasks had been covered for two of the engines. The studyteam wrote a number of task primitives on the whiteboard, to start theprocess. Then, the crew supervisors added new task primitives and reviewedand corrected the ones the study team had proposed.

To illustrate the process, skill and comprehension requirements for threeof the task primitives were identified by the crew supervisors.

When the same data were gathered for the three MAC tasks, no additionaltask primitives were required. The ones generated by the SAC tasks wereadequate to cover completely all the steps followed by the MAC technicians.However, another discussion was conducted to elicit the crew supervisors'opinions concerning an optimal training curriculum for engine shop tech-nicians. They produced the following training outline.

- Fundamentals of jet engine systems

-- Starter systems-- Fuel distribution and injection systems-- Pneumatic systems-- Electrical systems (generation, ignition, sensor)

2 - 3

- Hydraulic systems- Lubrication systems-- Mechanical systems (gearbox)- Jet engines (how the systems work together)-- Engine sections and what they do (compressor, turbine, diffuser,

combustion chamber)

General principles

-- Safety-- Proximity standards (jet, propeller)

--- Foreign object damage-- Wearing of jewelry

-- Tool use-- Accessibility (including use of stands)-- Lubricants (including the importance of proper disposal)- Technical orders (how to use)-- Forms and work unit codes

- Electrical systems

-- Safety-- Types of connectors- Reading schematics- Continuity checks-- Possible errors

- Use of tools and measuring equipment

-- Use/misuse-- Care of-- Accountability- Special tools

- Fuel systems

- Safety (grounding, flammability, fuel handling)- High/low pressure lines-- Contamination/filtering-- Color coding of lines

- Starting systems

-- Types (electrical, pneumatic, cartridge, jet fuel)- Safety (for each type)

- Ignition systems

-- Safety (grounding)-- Types-- Source of iqnition

2-4

- Lubrication systems

-- Types of lubricants- Safety

Use (how to apply)-- Handling (disposal, storage, contamination)-- Lubricants for hot and cold sections

- Hydraulic systems

- Types-- Safety-- High and low pressure lines

Data Analysis

The first step in data analysis was to consolidate the raw data that wereobtained concerning the steps, behavior objects, constraints, and errors foreach task. Removal and installation of the components were combined, and theindividual steps were collapsed is because within such groupings,task steps would likely require similar skills and comprehension and couldlogically be trained together. Appendix A presents the consolidated behaviorobjects, constraints, and errors for three remove-and-install tasks for threeweapon systems.

Nearly all the constraints found in this investigation represented designproblems that need to be solved on the current equipment and avoided in newengine designs. Some of the errors were task-specific, but many wereperfectly correlated with the presence of certain behavior objects. Forexample, technicians must avoid prolonged physical contact with petroleumproducts. A possible error when using a torque wrench is to overtorque orundertorque the connector(s), both possibly causing the connector to loosen orcome off. Similarly, standard errors may be associated with safety wire,gaskets, and electrical connectors. These standard errors and constraints canbe straightforwardly converted to specific design criteria for systemsincorporating these behavior objects.

The second step was to assign task primitives to the three tasks asperformed on the three engines. Note that since task primitives are definedin relation to specific behavior objects, and since those behavior objects inturn can be associated with weapon system specific design criteria, humanperformance of task primitives can be directly associated with the weaponsystem design. This is a prime motivation for using task primitives as afoundation for human performance analysis of yet-to-be-realized designs. (Itis anticipated that, similarly, typical comprehension and skill requirements,training requirements, and performance times may also be straightforwardlyassociated with eaf..h task primitive--both when these design criteria are metand when they are not. This would complete the connection of each proposeddesign approach with its representative MPT&S consequences. This assertionwill be tested in a future CAMT feasibility study.)

2-5

Results

Figure 5 lists our first and second attempts at task primitivedefinition. Figures 6 lists associated skill and comprehension requirementsfor three of these task primitives. Figure 7 shows the task applicability ofeach task primitive.

The objective of this field study was to demonstrate an ability to definea set of task primitives that are general, orthogonal, and complete. It isclear from examining these results that these criteria were substantially(albeit imperfectly) met.

Generalit.

Generality is important to the feasibility of a CAMT data base. To keepthe total number of task primitives in the data base well-scoped, each elementmust apply to many tasks on many weapon systems.

The task primitives first identified in this study are highly general;there was considerable overlap in the nine tasks under study (see Figure 7).Six of the task primitives appeared in all nine tasks. The task primitivesfirst identified for the SAC tasks were deemed adequate for describing the MACtasks. Of course, the tasks in this study were highly homogeneous. As othertypes of maintenance are investigated, new task primitives will have to beadded. However, many of the previously identified task primitives will carryover to tasks such as those performed by electricians, pneudraulictechnicians, egress technicians, environmental systems technicians, flightlinecrew chiefs, etc. Some will also be present in the tasks of electronic andavionics technicians.

Orthogonality

Orthogonality ensures there is never confusion about which of severaltask primitives to use when analyzing a segment of a maintenance task. Theoriginal list of task primitives constructed by the project team and crewsupervisors did not perfectly meet the orthogonality criterion. Four taskprimitives were deleted. The 27 initially identified task primitives became23, as shown in Figure 5.

Completeness

Completeness allows one to analyze any task into a set of task primitiveswith no segments left over. At first, this criterion seems trivial: onecould just define new task primitives to cover any leftover segments.Accommodating both completeness and generality, however, may be more of anart. Every task primitive should to apply to as many different systems andsystem designs as possible. (If task primitives themselves were defined asdesign-specific, they would no longer support extrapolation to new designs.There would soon be as many primitives as designs; i.e. a limitless set.)Therefore, completeness really implies that each segment of each task studiedshould be common to at least one other maintenance task-an intriguingassumption.

2-6

Z Z 0 -<0 r 0

C-4~4) C-4

-~06

w oS> 5,

.4

00

>1 w~0)

~Orco

6t3 ;j 0

U ~ ~ ').-'~. ~ cc

u~ ~p)

2-7

Task Primitive: Install/Remove Common Fasteners

1. Types 4. Application/seA. Nuts & bolts I. C-rings A. When to use - hot section, cold secuionB. Screws J. Roll pins B. Thread pitchC. Hairpin cottes C. Fixed/novable (a bolt with cotter pinD. Couplings/clamps is movable)E. Ball-lock pins (pip pins) D. Reusable/not reusableF. Cotter pins, ordinary E. Conservation of precious metalG. Rivets (hollow, solid, pop)H. Diaper pin cotters 5. How to remove hard-to-remove fasteners

A. Normal methods2. Installation & removal methods B. "Easy out"

C. Drilling3. How to avoid dissimilar metals D. Use of penetrating oil

Task Primitive: Apply Lubricants

1. Types 3. Application/UseA. Fuel D. Jellies A. How to apply (hand, brush, gun, etc.)B. Oil E. Dry film B. How much to useC. Grease F. Anti-siezc C. When

D. Where2. Hazards E. What types not to use under certain

A. Personnel conditionsB. EquipmentC. Spills & cleanup 4. Storage & disposal of lubricants

Task Primitive: Use Torue Wrench

I. Why torque wrenches are necessary 4. How to handle torque wrenchesA. How to read a calibration sticker

2. Types B. What to do if wrench is droppedA. Breakaway C. How to store wrenchB. Dial indicator D. How to zero out wrenchC. Ratchet

5. How to use torque wrenches3. Methods of setting A. Can it be used in both directions?

A. Pull down handle B. How to read the dial typeB. Locking collar C. How to calculate and set torqueC. Screw values when using crowsfoot,

extensions, and adaptersD. How to identify when the breakaway

type breaks away

E. False torques obtained when usingcrowsfot

Figure 6. Comprehension and Skill Requirementsfor Three Task Primitives

2-8

Task Primitive #

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 17 20 21 23 24 25 26 27

C-130F. Pump x x x x x xx x x x x

C-130 Starter x x x x x x x x x x x x x

C-130Turbine xx x x x x x x x x x x

KC-135F. Pump x x x x x x x x x x x x

KC-135Starter x x x x x x x x x x x

KC-135 Turbine x x x x x x x x

B-IBF. Pump x x x x x x x x x x x

B-IBStarter x x x x x x x x x x x x

B-IBTurbine x x x x x x x x x xx x x

Figure 7. Assignment of Task Primitives to Tasks

The 23 task primitives cover the training requirements for the nine tasksfairly completely. When the skill and knowledge requirements for all 23 taskselements are identified, the result will be a nearly complete andcomprehensive curriculum for engine shop incumbents. Figure 7 indicates somecause for concern, however, in that task primitives 4, 21, and 25 are eachused only once. Could they be design-specific? No. Their behavior objects--torque multipliers, push-pull fixtures, and use of heating/cooling forinstallation of tight-fitting parts-indicate potentially broad usage on manyother maintenance tasks.

To verify completeness, we compared the curricula that the SAC and MACsupervisors designed for engine shop technicians with the task primitiveswe identified together. We found several types of omissions in our taskprimitive list.

The first type consisted of curriculum topics that would have beenrepresented if all engine shop tasks, not just a selected few, had beenexamined. These include: powered and non-powered AGE, extraction of panelscrews, inspection for cracks/peeling/scores/etc., reading schematics, andcontinuity checks. These omissions are a result of the limited scope of ourstudy.

2-9

The second type of omission consisted of very common comprehensionrequirements: how to use technical orders, how to fill out maintenance forms,hazards associated with volatile liquids, and hazards associated withelectrical systems. These omissions would be contained in a Basic Skill andComprehension Requirements Dictionary, which would complement the TaskPrimitive Dictionary. They are not likely to be design-sensitive, but wouldbe important in predicting the overall training requirements of the system.

The third type of omission concerned the overall operation of the systembeing repaired: the sections of jet turbines and how they work together.There was some disagreement over exactly how much "theory" is required for ayoung two-striper, but there inevitably must be some. Again, the requirementto teach some general theory will not be design- sensitive. If needed forpredicting training resources, however, it could be covered in the BasicSkills and Comprehension Requirements Dictionary.

The final type of omission consisted of requirements dealing withspecific behavioral objects: fuel contamination, oil filtering, color codingof fuel lines, etc. These would be cove-red with more specific definitions oftask primitives (see discussion below).

Within the limited scope of this initial feasibility study, we candeclare our goals of defining task primitives that are general, orthogonal,and complete to be successful. This, however, is just one of severalfeasibility issues fundamental to CANT design analysis.

Discussion

The primary use of task primitives is to bridge design features withtheir consequent MPT&S requirements. Defining task primitives toospecifically-too close to the task-step level-risks losing their generality,requiring too many of them for a practical data base, and losing theirconnection to MPT&S requirements. Defining task primitives too broadly- tooclose to whole-task level--risks losing their ability to tie specific humanperformance to specific design configurations.

As discussed in the concept description, a CANT analysis could havesignificant impact in several major areas: providing immediate designfeedback, predicting MPT&S requirements once the design is set, andrestructuring training more efficiently. When describing our study to theline chiefs, we stressed the latter, since this was a topic of constantconcern for them. Manpower policy and acquisition problems were far lessfamiliar.

Our first attempt to define task primitives, therefore, particularlystressed grouping skill and comprehension requirements which could be trainedtogether for maximum efficiency. It gave little consideration to designutility. In retrospect, some primitives are defined too broadly forpredicting the human performance consequences of design variations.

2 - 10

Consider:

6. Clean components11. Remove and install heavy components14. Use physical measuring devices15. Inspect components for possible reuse27. Use common hand tools

Could one find three representative design configurations for each whichwould scope most human performance consequences? Could one measure typicaltimes to perform each? No, they each cover too many different cases.

These primitives will work, however, if we increase the specificity oftheir behavior objects. Referring to Figure 6, we could split "Use TorqueWrench" into two task primitives: one for breakaway wrenches and one for dialindicator wrenches. We could similarly define a separate task primitive foreach type of common fastener: nuts and bolts, rivets, C-rings, etc. We couldsimilarly define a separate task primitive for each type of lubricant: fuel,jellies, dry film, etc.

Task primitives defined at this level of specificity would retain a highdegree of generality, orthogonality, and completeness. Yet, by addressingbehavior objects of more specific design configuration, they could morespecifically address the design of the behavior objects' immediateenvironment. Also, their human performance consequences could more easily bemeasured via parametric studies or in the field.

Several related task primitives (e.g., the two for torque wrenches) mightindeed share basic skill and comprehension requirements, and could efficientlybe trained together in a set. This is compatible with the CANT concept aslong as they remain orthogonal. Furthermore, some task primitives in Figure 5might be better defined as basic skill and comprehension requirementsthemselves: remove and install heavy objects, and dispose of waste fluids.Such basic skill and comprehension requirements would only differ betweendesign approaches if the differences in the approaches themselves were major(say, using different technologies). Usually, general basic skill andcomprehension requirements will not have design consequences, but areimportant for scoping the overall skill and training requirements of (sub)system maintenance.

As we explore the building of the accompanying CAMT data base, we willcontinually fine-tune the task primitive definitions. Overall, our firstattempt to define primitives of real-world maintenance tasks supports theplausibility, at least, of bridging design consequences with MPT&Srequirements.

Future studies must address the following issues:

- Could a well-scoped set of general, orthogonal, and complete taskprimitives be further defined for electronic (versus mechanical)systems maintenance? For diagnostic/inspection (versusremove/replace) tasks?

2 - 11

Could the Task Primitive Dictionary and Comprehension and BasicSkill Requirements Dictionary be populated with detailed,credible data in a relatively efficient manner? Could one arriveat a predicted cost per specified number of task primitives tobuild the data base in order to predict the cost of a completeCAMT system?

Could CAMT analysis be demonstrated to perform credible designanalysis relative to MPT&S requirements, given an appropriatesample problem?

2 - 12

SECTION III. ANNOTATED BIBLIOGRAPHY

Introduction

The Air Force tasked the contractor to conduct a literature review toobtain a theoretical foundation for task-primitive and skill definition. Thespecific purposes of the literature review were:

- To determine whether an "anatomic" approach similar to CAMT hasbeen attempted before. If it has, to determine the reasons forits success or failure. If it hasn't, to highlight how othertaxonomies compare with CAMT.

- To gain insight into methods for deriving general, orthogonal,and complete task primitives fLom a set of maintenance tasks.

- To gain insight into efficient means for populating the taskprimitive and skill-and-comprehension-requirements data bases,once the task primitives have been defined.

The Air Force performed a key-word search of the Defense TechnicalInformation Center (DTIC) data base to obtain a listing of possibly relatedliterature. The Air Force then used DTIC abstracts to select documents to bereviewed by the contractor with respect to the above objectives. Alsoincluded were some references personally accumulated by Capt Donald R. Loose.The contractor summarized each assigned document and noted relevance to CANT.The contractor included negative findings; i.e., when relevant-soundingabstracts led to irrelevant documents. The findings follow.

Bibliography

Alley, W. E., Treat, B. R., and Black, D. E. (1988, September). Classificationof Air Force Jobs into Aptitude Clusters (AFHRL-TR-88-14). Brooks AFB,TX: Air Force Human Resources Laboratory, Information Sciences Division.(AD A2066610)

The Air Force has four job clusters: Mechanical, Administrative,General, and Electronics. This study applied a new procedure for homogeneousclustering of regression equations in an Air Force ASVAB validity studyinvolving 155,000 recruits in 211 technical training programs.

The present system of forming composites into the four job clusterslisted above is remarkably robust, considering the myriad of changes that havetaken place since the system was first established. A number of specialtieswere not well predicted--some because they have few cognitive demands, othersbecause the demands they make are not sufficently represented in the currentASVAB. A recommendation was made to study these specialties further.

3-I

A recommendation was also made to consider forming a newcluster/composite. The tactical/strategic aircraft maintenance specialtiesincluded in this group seemed to reflect a "generalist" requirement-one thatrequired abilities across the full domain of subtest measures.

While overall performance within each of the four job clusters correlateswell with specific ASVAB profiles, it is not clear what proportion of the taskprimitives comprising each cluster would share this correlation. Until theeffects of ASVAB scores on maintenance performance are studied on a primitive-by-primitive basis, they will be of no aid in a CAMT analysis.

Driskill, W. E., Weissmuller, J. J., Hageman, D. C., and Barrett, L. E.(1989, August). Identification and Evaluation of Methods to DetermineAbility Requirements for Air Force Occupational Specialties(AFHRL-TP-89-34). Brooks AFB, TX: Air Force Human Resources Laboratory,Manpower & Personnel Division.

This project reviewed the literature and identified 36 taxonomies that havebeen applied to the description of job and/or worker characteristics. The majororientation of this study was personnel selection. The authors were looking forjob descriptive methods that can be used to derive aptitude and abilityrequirements. The following seven methods were chosen for further evaluation:Functional Job Analysis, Job Element Method, Position Analysis Questionnaire,Occupation Analysis Inventory, General Work Inventory, Threshold Traits AnalysisSystem, and Ability Requirements Scales.

The authors concluded that the Ability Requirements Scales, developed byFleishman and his coworkers, was the most appropriate of the existing methodsfor identifying the ability requirements of various Air Force occupationalspecialties. However, they found that no taxonomy adequately covered social andinterpersonal communication abilities. They also felt that certain cognitiveabilities that can be measured only through computer test administration havebeen omitted from existing taxonomies (e.g., the ability to visualize movingspatial objects in various configurations and at differing speeds).

They recommended that a new Air Force taxonomy of occupational abilities bedeveloped. "The ability requirements for an occupational specialty would bedetermined through subject-matter-expert ratings of ability requirements fortask statements categorized by the verbs in the task statements. This approachwould enable generalization across occupational specialties. If this methodwere applied and ratings established for all relevant tasks, ability requirementprofiles could be generated for any Air Force occupational specialty without theexpense of subject-matter-expert conferences" (p. ii).

The latter notion of utilizing the results of previous analyses to savetime and effort is also a feature of CAMT. Driskill et al. would use theratings established for all relevant verbs to generate ability requirementprofiles for new specialties. CAMT will use the data gathered from existingtasks in an existing weapon system to assist in analyzing the tasks in new

3-2

weapon systems. Where the two systems have task primitives in common, thedata from the older system will be transported to the new system.

Drury, C. G., Paramore, B., Van Cott, H. P., Grey, S. M., and Corlett, E. N.(1987). Task Analysis. In G. Salvendy (Ed.), Handbook of Human Factors(pp. 370-401). New York: John Wiley & Sons.

This chapter explores the origins and antecedents of task analysis andjob analysis. These developments are described in some detail to show thatthere is no single method of task analysis applicable to all jobs. Finally, atask analysis of a large, complex system is presented in detail to illustrateboth the techniques of task analysis and the methods of collecting theinformation required.

One of the methods described is the Position Analysis Questionnaire (PAO)of E. J. McCormick. It has some characteristics in common with the techniquesof the current CAMT project. The authors state, "The PAQ is clearly a dif-ferent type of job analysis instrument from typical task analysis instrumentsin that it provides for the analysis of jobs in terms of basic human behaviorsthat cut across various types of jobs, rather than in terms of specific tasksthat characterize the work activities of particular jobs or small groups ofjobs" (p. 383). The CAMT technique also involves analyzing tasks to detectthe presence of any of a previously compiled set of human behaviors (taskprimitives).

The PAQ asks 189 questions, with each question requiring a rating.Nearly all of the rating scales have five points. The questions are groupedinto six major categories: information input, mediation processes, workoutput, interpersonal activities, work situation and job context, andmiscellaneous aspects. The primary uses of the PAQ have been: personnelselection, personnel development and training, career ladder development,performance appraisal, and job evaluation (establishing rates of pay).

The task analysis methods reviewed in this chapter have the followingcharacteristics in common with CAMT.

- They examine the activities of tasks.

- They identify their performance demands.

- They identify the skills and knowledge needed to perform them.

- They can be used to plan training curricula and course content.

CAMT and the typical method reviewed in this chapter differ in thefollowing ways.

- CAMT stops short of describing tasks at the step level becausethe decisions that CAMT is designed to support (MPT&S decisions)do not require step-level description and analysis.

3-3

- CAMT is more efficient in that task primitives found inpreviously described tasks can be used to describe new tasks, asapplicable.

Eggemeier, F. T., Fisk, A. D., Robbins, D., Lawless, M. T., and Spaeth, R. L.(1988, November). High Performance Skills Task Analysis Methodology: AnAutomatic Human Information Processing Theory Approach (AFHRL-TP-88-32).Brooks AFB, TX: Air Force Human Resources Laboratory, Logistics andHuman Factors Division. (AD B128366)

The study described in this report attempted to determine whetherpart-task trainers aimed at the psychological facility of "automaticity" wouldbe effective. "Automaticity" is a mode of information processing andresultant reaction that is largely automatic and seemingly effortless, butwhich handles complex and voluminous information. The focus of the study wason TAC and Air Force Space Command comand and control tasks.

A distinction was made between two qualitatively different forms ofinformation processing used by the command and control system operators:automatic and controlled. Automatic processing develops with extendedpractice under consistently mapped conditions for which there is a consistentrelationship among task components. The following examples of automaticprocessing when driving a car were cited: applying the brake, interpretingtraffic lights, and shifting gears. Controlled processing is typicallyassociated either with novel tasks or variably mapped conditions for whichtask component relationships vary from situation to situation. Examples ofcontrolled processing when driving a car are: interpreting a new road map,planning the route to be taken, and increasing the separation between vehiclesto compensate for bad road conditions.

Observations and interviews with controllers enabled the study team todecompose controllers' tasks into automatic and controlled. The studyconcluded that much of an experienced controller's proficiency involvesautomatic processing of consistently mapped tasks, and that these tasks couldbe taught on specialized part-task trainers, thereby reducing training timeand cost.

The relevant implication for the CAMT effort is that there exist tasks orportions of tasks that improve little with practice (where there are variablymapped conditions). If this is true, it may be because the necessary skillshave already been acquired, not because the stimuli are variable. People whoare learning to drive improve little in reading a road map only when they arealready highly skilled in doing it. When they have a lot to learn (e.g.,interpretation of map symbols), they improve with practice. CAMT notes asimilar distinction in the definition of basic-skill versus practiced-skillrequirements.

3-4

Erickson, R. A. (1986, September). Measures of Effectiveness in SystemsAnalysis and Human Factors (NWC TP 6740). China Lake, CA: Naval WeaponsCenter. (AD A175353)

This report presents the thesis that the measures of effectiveness (MOEs)of any system are hierarchical in nature, ranging from the top-level MOE thatindicates the worth of a system, through MOEs of mission capability, down toMOEs (or measures of performance) of individual tasks making up a mission.The top-level hierarchy includes factors such as cost, survivability, reli-ability, and capability. Guidelines are presented for developing the MOEs fora system's capability to accomplish a mission.

An MOE may be defined as a measure of the extent to which a system can be

expected to complete its assigned mission within an established time frameunder stated environmental conditions.

This study has little practical relevance to CAMT because its methodologyoperated at a higher level of system description than does CAMT's. Ericksonfocused on the identification of system objectives, the functions to beperformed in attaining these objectives, and possible measures for evaluatingtheir accomplishment. No attempt was made to describe or analyze the specificcharacteristics of tasks.

Fleishman, E. A. and Quaintance, M. K. (1984). Taxonomies of Human Perform-

ance. Orlando, FL: Academic Press.

This book "describes and evaluates some taxonomic efforts to developclassificatory systems for describing human task performance. The objectivesof such systems, their role in scientific development, and the practicalimplications are discussed" (p. 1).

Fleishman (1967) distinguishes abilities from skills. An ability refersto a more general capacity of the individual related to performance in avariety of human tasks. Fleishman (1967, p. 352) stated, "The fact thatindividuals who do well on task A also do well in tasks B and C but not intasks D, E, and F indicates, inferentially, a comon process involved inperforming the first three tasks distinct from the processes involved in the

latter three. To account for the observed consistencies, an ability ispostulated." Thus, an ability is a general trait of the individual that hasbeen inferred from certain response consistencies. Both learning and geneticcomponents underlie ability development. In contrast, a skill is defined asthe level of proficiency on a specific task or group of tasks. The develop-ment of a given skill or proficiency on a given task is predicated, in part,on the possession of relevant basic abilities (pp. 162-163).

Under the CAMT methodology being developed, skill will have a completelydifferent definition. A skill will be a task requirement that demandspractice for its acquisition by the task performer.

3 - 5

In defining ability factors, Fleishman and his colleagues are reallylinking together information about task characteristics with ability require-ments. One may say that a person possesses the ability or, alternatively,that the task requires, involves, or elicits the use of the ability.

While Fleishman has tried many ways to classify human performance, themajor one presented in this book is based on what human beings can do-ongeneral abilities that have been defined through psychological studies. Bycontrast, the CAMT classificatory approach is based on what maintenanceworkers actually do--on task primitives derived from observation of workperformance and interviews with workers.

A study described in detail is one by Theologus, Romashko, and Fleishman(1973). They developed an Ability Requirements Approach, which requiredjudges to rate tasks on 37 ability dimensions, using a seven-point scale foreach ability. The human abilities definitions are listed below (abbreviatedfrom pp. 322-326).

1. Verbal Comprehension - Understanding words and words in context.

2. Verbal Expression - Communicating ideas or facts. Unrelated to thequality of the ideas.

3. Ideational Fluency - Producing ideas concerning a given topic.Number, not quality, of ideas.

4. Originality - Producing unusual, clever, creative responses. Degreeof creativity, not number.

5. Memorization - Recalling accurately information presented during thetask. Not recalling task procedures.

6. Problem Sensitivity - Recognizing the whole problem and all of itselements. Not reasoning.

7. Mathematical Reasoning - Understanding or structuring of math prob-lems. Not manipulation of numbers.

8. Number Facility - Speed and accuracy of computation.

9. Deductive Reasoning - Applying general concepts or rules to specificases. Involves the ability to synthesize disparate facts into gen-eral principles.

10. Inductive Reasoning - Finding general concepts or rules to explainobserved relationships.

11. Information Ordering - Applying rules or objectives to arrangeinformation into the most appropriate sequence.

12. Category Flexibility - Ability to produce alternative groupings orcategorizations for a set of items.

3 - 6

13. Spatial Orientation - Maintaining one's orientation, in relation toobjects in space or comprehending where objects in space are, inrelation to observer's position.

14. Visualization - Manipulating or transforming visual images ofspatial patterns. Visualizing how they would appear after specifiedchanges.

15. Speed of Closure - Speed of organizing apparently disparate elementsinto a single meaningful pattern or configuration. All elements arein the same sensory modality.

16. Flexibility of Closure - Detecting a previously specified stimulusconfiguration among distracting stimuli. The relevant and distract-ing stimuli are in the same sense modality.

17. Selective Attention - Performing a task despite distractions fromoutside the task or monotonous conditions.

18. Time Sharing - Utilizing information coming from two or morechannels of communication. May integrate the information and use itas a whole, or retain and use it separately.

19. Perceptual Speed - Speed of comparing sensory patterns to determineidentity or degree of similarity. Patterns may be presented succes-sively or simultaneously. The comparisons may be between rememberedand presented patterns.

20. Static Strength - Applying continuous effort to lift, push, or pulla fairly immovable or heavy external object.

21. Explosive Strength - Expending a burst of muscular effort, as injumping, sprinting, or throwing.

22. Dynamic Strength - Repeatedly or continuously supporting or movingthe body's own weight by using the power of arm and trunk muscles.

23. Stamina - Maintaining physical activity over prolonged periods oftime. The resistance of the cardiovascular system to breakdown.

24. Extent Flexibility - Extending, flexing, or stretching musclegroups. Degree of flexibility. Not repeated or speed flexing.

25. Dynamic Flexibility - Repeated trunk and/or limb flexing. Bothspeed and flexibility required. Ability of muscles to recover fromflexing.

26. Gross Body Equilibrium - Balancing the body; not balancing objects.

27. Response Orientation - Speed of initiating appropriate response,given a stimulus, where two or more stimuli and two or moreresponses are possible. Not speed with which response is carriedout.

3-7

28. Reaction Time - Speed of initiating a single motor response to asingle stimulus.

29. Speed of Limb Movement - Speed of moving arms and legs. Not speedof initiation of movement. Not precision, accuracy, orcoordination.

30. Wrist-Finger Speed - Speed of moving wrist and fingers. Same as 29above.

31. Gross Body Coordination - Coordinating movements of trunk and limbs.

32. Multilimb Coordination - Coordinating the movement of two or morelimbs, with the body at rest.

33. Finger Dexterity - Making skillful coordinated movements of fingers.Manipulation of objects may or may not be involved. Notmanipulation of control mechanisms. Not speed of movement.

34. Manual Dexterity - Making skillful, coordinated movements of handsor hands and arm. May include manipulation of objects, but notmanipulation of control mechanisms.

35. Arm-Hand Steadiness - Minimizing tremor and drift while maintaininga static arm position. Not manipulation of control mechanisms.

36. Rate Control - Making motor adjustments to intercept or follow acontinuously moving stimulus whose speed and/or direction vary in anunpredictable fashion. Not where they are predictable.

37. Control Precision - Adjusting or positioning controls. Can be anti-cipatory motor movements in response to changes whose speed anddirection are perfectly predictable.

Hagman, J. D. (1980, May). Effects of Training Task Repetition on Retentionand Transfer of Maintenance Skill (Research Report 1271). Alexandria,VA: U.S. Army Research Institute for the Behavioral and Social Sciences.(AD A101859)

This report describes a transfer-of-training study which examines theeffects of practice with a 500A Sun Test Stand. It examined whether practicewith testing of a 100-ampere alternator transferred to testing of a 60-amperegenerator. The study had nothing to do with task taxonomies or task analysis.

Hosek, J. R. and Peterson, C. E. (1988, November). Developing an InitialSkill Training Database: Rationale and Content (N-2675-FMP). SantaMonica, CA: The Rand Corporation. Prepared for The Office of theAssistant Secretary of Defense for Force Management and Personnel,Contract No. MDA903-85-C-0030. (AD A201689)

3-8

In 1983, the Secretary of Defense created the Training and PerformanceData Center (TPDC). It includes information on training equipment andteaching aids such as simulators. Data on individual job performance and unitproductivity were added as they became available. TPDC asked Rand Corporationto describe the kind of data that would be particularly useful for the initialskill training (IST) of individuals.

They concluded that data should be gathered on aspects of personneltraining that relate to or affect military capability. Specifically, the ISTdata base files would contain data pertaining to: individuals, courses,occupation, units, and training resources. The data would be applied in thefollowing five areas.

- Determining training loads for budgeting.

- Matching weapons system acquisition and training requirements.

- Managing skills shortages.

- Selecting personnel for military occupations.

- Determining the value of military training to the individual.

The report recommends that the TPDC gather course-descriptive data in itsIST data base files (e.g., data on course content, skills taught, and prereq-uisites). If the course-content (curriculum) data were to include the timedevoted to each course topic, these data could be used to flesh out the CAMTBasic Skill and Comprehension Requirements Dictionary. Unfortunately, it isnot clear from the report that time-per-topic data is to be gathered. Aspresented, the most useful data for CAMT will be the skills taught in eachcourse. These data can provide CAMT investigators leads for identifyingsources they need.

Kraiger, K. (1989, April). Generalizability Theory: An Assessment of itsRelevance to the Air Force Job Performance Measurement Project(AFHRL-TP-87-70). Brooks AFB, TX: Air Force Human Resources Laboratory,Training Systems Division. (AD A207107)

This paper reviews the applicability of generalizability theory to theAir Force's Job Performance Measurement Project. The theory is illustrated byapplying it to data collected from Air Force jet engine mechanics.

Generalizability theory is a method for estimating the dependability ofscores over various conditions of measurement. In contrast to classical testtheory (which permits the investigation of only one error source at a time),generalizability theory allows the researcher to simultaneously investigatemultiple sources of error.

Classical test theory provides a single, true score but multiple esti-mates of true-score variance, depending on how error variance is defined.

3-9

Generalizability theory, on the other hand, explicitly recognizes the exist-ence of multiple sources of error variance (such as items, test occasions, andraters) and provides methods for simultaneously estimating each.

Proficiency ratings were collected for 256 first-term jet enginemechanics. There were three facets of generalization: rating forms (4),specific items on each form (2 to 32), and rating sources (self, peer, andsupervisor).

The question of interest was whether performance scores weregeneralizable (consistent) over different rating sources, forms, and iteas.The results indicated that scores were generalizable over both forms and itemswithin forms. However, scores were not generalizable over rating sources.Sources tended to differentially rank ratees, depending on the specific formused. This finding suggested that it may be inappropriate to average scoresover sources and that separate analyses for other facets should be conductedwithin rating sources.

The relationship of generalizability theory to construct validity and thelogical requirements for performance ratings are discussed.

This paper extends generalizability theory-a tests and measurementtheory--to job performance rating. It has no relevance to the CAMTmethodology.

Lintz, L. M., Loy, S. L., Brock, G. R., and Potempa, K. W. (1973, August).Predicting Maintenance Task Difficulty and Personnel Skill RequirementsBased On Design Parameters of Avionics Subsystems (AFHRL-TR-72-75).Brooks AFB, TX: Air Force Human Resources Laboratory, Advanced SystemsDivision. (AD 768415)

The study had two primary goals:

- to investigate the relationships among equipment designcharacteristics, task difficulty, task performance time, anderror probability; and

- to investigate the relationships between performance variablesand personnel variables.

Organizational-level maintenance data were collected on 27 functionalloops from 10 avionics systems. (A functional loop was defined as a networkof circuits and equipment that performs a specific function.)Intermediate-level maintenance data were collected on 28 line replaceableunits (LRUs) for the same 10 avionics systems.

Twenty-eight (29 for intermediate-level maintenance) equipment designcharacteristics were either measured objectively or rated by AF personnel.Data on 16 personnel characteristics were obtained.

3 - 10

Three maintenance tasks were identified for each functional loop andLRU-a functional checkout task, an easy task, and a difficult task. For eachtask, experienced AF maintenance supervisors estimated performance times anderror probabilities for their high-skill and low-skill personnel. Supervisorsalso rated the difficulty of each task.

The design, personnel, performance, and difficulty variables wereintercorrelated, and entered into regression and factor analyses. Equationswere derived to predict performance and task difficulty from designcharacteristics, and to predict performance from personnel variables and taskdifficulty. Personnel factors and design factors were identified. Personnelprofiles were developed for high- and low-skill groups. Plots of performanceversus difficulty and task completion versus time were developed.

At the organizational level, six performance and personnel factors thatwere isolated were: Aptitude, Experience, Motivation, Breadth of Skills, AirForce Technical Training, and Time in Grade. Intermediate factors were:Length of Service, Aptitude, Electronics Aptitude, Avionics Experience, andNon-Air Force Technical Training.

Factor analysis of design variables indicated that factors associatedwith performance time and errors were: Checkout Complexity, CheckoutInformation, Length of Checkout, Accessibility, Equipment Complexity, and TestEquipment and Adjustments.

At organizational and intermediate levels, the low-skill groups requiredapproximately twice the performance time of the high-skill groups, and hadapproximately six times the error probability. Aptitude was related to timeand errors at the organizational level but not at the intermediate level,perhaps because the selection process had created greater homogeneity ofaptitude for the intermediate personnel.

The relationships that were found are not unexpected. They reinforce thenotions that:

- people with higher levels of aptitude, experience, and trainingtend to perform maintenance tasks more quickly and with fewererrors, according to maintenance supervisors; and

- equipment with lower levels of complexity, checkout complexity,and checkout length--and higher levels of checkout completenessand accessibility--is easier to maintain, as shown in experts'ratings.

The 2:1 variation in performance time and 5:1 variation in the number oferrors between low-skilled and high-skilled maintainers supplies a useful rulefor performing P, T, and S tradeoffs, but not for predicting design effects.

3 - 11

Lobel, A. E. and Mulligan, J. F. (1980, January). Maintenance TaskIdentification and Analysis: Organizational and Intermediate intenance(AFHRL-TR-79-50). Brooks AFB, TX: Air Force Human Resources Ip 'atory,Advanced Systems Division. (AD A083685)

This report provides a draft specification that sets forth a process forobtaining the task-descriptive information to be incorporated into job-guidemanuals and logic-tree troubleshooting aids. It speaks of "maintenance taskanalysis," which means "the analysis of the identified task to determine whatthe task consists of, what is needed to perform it, and how it should beperformed" (p. 9).

The process includes the development of:

- a task identification matrix,

- a description of the intended user,

- listings of required support equipment and special tools,

- guidelines for determining the level of detail for writingjob-guide manuals and logic-tree troubleshooting aids,

- an analysis of possible equipment faults and the resultingsymptoms,

- effective step-by-step procedures for accomplishing each task,and

- action trees outlining a troubleshooting strategy for isolatingeach possible fault.

The methodology described in this specification addresses the developmentof performance aids after detailed design is accomplished. It is unrelated toCAMT.

Perrin, B. M., Knight, J. R., Mitchell, J. L., Vaughan, D. S., and Yadrick,R. M. (1988, September). Training Decision System: Development of theTask Characteristic System (AFHRL-TR-88-15). Brooks AFB, TX: Air ForceHuman Resources Laboratory, Training Systems Division. (AD A199094)

The Training Decisions System (TDS) is a computer-assisted decisionsystem that aids in planning the what, where, and when of training for AirForce career ladders. One of four basic subsystems of the TDS is the TaskCharacteristics System. The Task Characteristics System addresses what tasksneed to be trained and where to conduct the training (training settings). Ithas two components:

- to construct Task Training Modules (Tr's) from occupationalsurvey data; and

3 - 12

- to allocate the TTMs to the various training settings (such asformal training, OJT, etc.), and to determine the number of hoursof training in each setting that is required, on average, toreach minimum performance standards for each TTM.

Both components rely heavily on the opinions of subject matter experts(SMEs). First, the tasks of an AFS were clustered statistically, usingcopeLformance as the similarity measure. (Coperformance is defined asfollows: given that an airman performs one task, coperformance is the proba-bility that he or she performs other specific tasks.) Next, separate teams ofSMEs were asked to cluster the tasks into groups that should be trainedtogether. The SME teams then met to reconcile their differences and produce afinal set of TTMs. Finally, for each TTM, the SMEs were asked to indicate thetime they believed it would take to completely train the tasks to miniuimstandards, given a specific type of training. The types of training were:classroom instruction; correspondence courses; hands-on experience in small,supervised training groups; and hands-on experience on the job. If the taskscould not be completely trained using a particular type of training, the SMEsindicated the percentage of full training that can be provided and the time itwould take to reach that level of proficiency.

After applying the methodology to four AFSs, the authors concluded thattheir survey procedure could yield adequately stable estimates of proficiencygains to be derived from hours of specified types of training.

If TTMs could be used as a basis for task primitive definition, the TDSdata base could populate much of the CAMT data base. In CAMT, expertmaintenance training personnel will be asked to estimate the likelihood of theerrors that are listed and to rate the seriousness of their consequences, aswell as to estimate the time to acquire specific skills and knowledge, and theexpected frequency of refresher training.

Primoff, E. S., and Eyde, L. D. (1988). Job Element Analysis. In S. Gael(Ed.), The Job Analysis Handbook for Business, Industry, and Government(Vol. 2). New York: John Wiley & Sons.

This chapter describes the Job Element Method (JEM) of job analysis, atechnique for recruitment, selection, performance rating, promotion, andtraining design that the U.S. Office of Personnel Management has researched.Expert workers and their supervisors are asked to identify significantelements in a job and rate them. These may be intellectual elements, motorelements, or work habits. Elements are identified by asking whatcharacteristics make a worker superior and what characteristics make for aweak worker. The elements are analyzed into subelements, or specific evidenceof the presence or absence of an element. The subelements are rated to showthe degree to which each subelement represents marginal behavior, superiorbehavior, behavior likely to cause trouble if not corrected, and behavior thatis sufficiently prevalent among job applicants to be practical to consider.Primoff defines "job element" as a worker characteristic that influencessuccess in a job. An element may be "a knowledge, such as knowledge of

3 - 13

accounting principles; a skill, such as skill with woodworking tools; anability, such as ability to manage a program; a willingness, such as willing-ness to do simple tasks repetitively; an interest, such as interest inlearning new techniques; or a personal characteristic, such as reliability ordependability" (Primoff, 1975, p. 2).

The job elements of JEM are clearly at a higher (less detailed) level ofdescription than the task primitives of CAMT. For example, where the JEM issatisfied with the job element, "skill with woodworking tools," CANT wouldhave a separate task primitive for each of the woodworking tools. Anotherdifference between Primoff's "job element" and CAMT's "task primitive" is thatthe job element is an attribute of a worker, while a task primitive is anattribute of the task. The task primitive is a set of behaviors that would betaught together when the task is trained.

Ramsey-Klee, D. M. (1979, December). Taxonomic Approaches to EnlistedOccupational Classification: Volumes I & II (NPRDC TR 80-7). San Diego:Navy Personnel Research and Development Center. (AD A078667 & ADA122028)

This study examines the taxonomic structure underlying the design of theNavy Occupational Task Analysis Program (NOTAP) task inventory booklets. Itpresents two alternative taxonomic procedures that will shorten task inven-tories and extend the usefulness of task analysis data.

A 55-item literature survey of taxonomic and classificatory methodologyand systems is presented in Volume II and summarized in Volume I. Silverman's(1967) definition of the taxonomic process is as follows:

A taxonomy involves the systematic differentiation, ordering,relating, and naming of type groups within a subject field.... Thetaxonomic process involves the following steps.

1. Collecting samples of phenomena.

2. Describing essential features or elements.

3. Comparing phenomena for similarities and differences.

4. Developing a set of principles governing the choice andrelative importance of elements.

5. Grouping phenomena on the basis of essential elements intoincreasingly exclusive categories and naming the categories.

6. Developing keys and devices as a means of recognizing andidentifying phenomena.

This study includes a statistical analysis of task-inventory responsedata for five Navy enlisted ratings and a content analysis of NOTAP task

3 - 14

inventory booklets. Essentially, the study was aimed at reorganizing andshortening the NOTAP task inventory booklets used by the Navy in job analysis.The author recommended deleting job titles and reorganizing the taskstatements around a new set of categories developed during the project.

one of the statistics calculated was a frequency rank ordering of theequipment, tools, systems, and supplies that incumbents in each of the ratingsuse, operate, or repair. These lists contain items similar to the "behaviorobjects" CAMT analysts will consider when naming task primitives. Except forthis similarity, the methods of the study and the taxonomic structure used toreorganize the task statements bear little resemblance to CAMT.

Reigluth, C., Merrill, M. D., Branson, R. K., Begland, R., and Tarr, R. (1980,August). Extended Task Analysis Proceure (ETAP) User's Manual. FortMonroe, VA: Training Developments Institute. (AD A098351)

This report was written as a reference document for the analyst who hasalready completed training in the use of Extended Task Analysis Procedure(ETAP). ETAP is a 12-step process designed to analyze (1) tasks that areprimarily procedural in nature and (2) tasks in which the soldier has to adaptto current situations ("soft skill" tasks).

ETAP has three phases: process analysis, substep analysis, and knowledgeanalysis.

- Process Analysis. Preparing a flow chart that shows each of thetask steps in proper sequence--identifying the decision steps andalternative sequences of steps (branches) which result from thedecision.

- Substep Analysis. Applying process-analysis methodology to eachstep identified in the process analysis. This more detailedanalysis is repeated until the resulting step is in terms ofskills which the student has already acquired prior totraining.

- Knowledge Analysis. Not all prerequisite skills or knowledge arebest represented as procedures. Some involve knowing facts orpieces of data. Some are concepts and require the individual toidentify objects or events in terms of their class membership.The process is to ask about each step or substep identified inphases 1 or 2: "What does the student need to know to performthis step?" If the answer is a concept, the same question isasked about that concept. Knowledge analysis is repeated untileach identified concept is one already mastered by the studentprior to instruction.

Some tasks require a different procedure each time they are accomplishedbecause of constraints, conditions, and circumstances ("factors") that vary ineach situation. The authors call such tasks "soft skills" or "transfer

3 - 15

tasks." In such tasks, the task performers learn the principles underlyingperformance rather than a specific procedure. Each time the task isperformed, the soldier derives a new procedure that is appropriate to thecurrent set of circumstances. The training for such a task involves teachingthe soldier (1) to recognize the various factors in a situation and (2) toderive procedures appropriate to each constellation of factors. ETAP has twoanalysis procedures to handle such tasks: factor-transfer analysis andprinciple-transfer analysis.

The methodology described in this manual results in a much more detaileddescription of tasks than CAMT provides for. It is truly "extended." For aprocedural task, all the steps are listed--in correct procedural order. Ifany step is described at such a general level that the lowest-level enteringsoldier would not be able to understand what is to be done, they are brokendown into substeps. For each "decision" step, the analyst identifies allfactors that need to be considered when making this decision. This level ofanalysis is useful for the development of performance aids and preparation oftraining based on those performance aids. However, it is too detailed formaking the types of decisions for which CAMT is designed (MPT&S decisions).Some methods for identifying knowledge and skill requirements could possiblybe adapted for CAMT. However, their concept of "basic skills" includes"steps, facts, concepts, and principles" (see pp. 84 & 107); CAMT's does not.

The manual is well organized and probably quite helpful to the analyst.The total ETAP process is described and flow-charted. Checklists guideanalysts through the process. For each portion of the process, there is asample dialogue between an analyst and an SME. This dialogue shows the kindsof questions that need to be asked and the types of answers that can beexpected. The analyst frequently recaps the SME's reply to test his/herunderstanding of the response.

Root, R. T. (1957, March). Annotated Bibliography of Research Studies inAviation Mechanical Maintenance. The George Washington University, HumanResources Research Office, Staff Memorandum. (AD F630578)

This memorandum describes reports concerned only with non-electronicmaintenance. The first section (Mechanic Evaluation) describes and evaluatesthe graduates of ten mechanic courses given at Air Proving Grounds, Eglin AirForce Base, FL. The second section is intended to describe studies in taskanalysis, proficiency measurement, and criteria measurement. However, thesestudies are more similar to occupational measurement or job analysis than taskanalysis as we now think of it. They did not address skills, knowledges, orphysical demands. The third section sets forth the "objective and approach"of seven projects being performed at the Maintenance Laboratory, Air ForcePersonnel and Training Research Center, Denver, CO. They deal with the designof job aids, training aids, and methods for identifying the skills, technicalinformation, and concepts to be covered by job aids and training aids.

3 - 16

Ruck, H. W. (1983, March). An Analysis of Selected Electronic Specialties forSkill/Knowledge Commonality. Brooks AFB, TX: Air Force Human ResourcesLaboratory.

This report describes a methodology for selecting electronic specialtiesfor consolidation. The purpose of the consolidation is to effect moreefficient and effective personnel management, personnel utilization, andtechnical training. The consolidation of specialties offers the followingspecific advantages:

- increased operational flexibility in utilizing personnel withinfield units;

- simpler assignments due to larger pools of eligible incumbentsand fewer specialties;

- fewer initial-skills courses, resulting in simpler trainingmanagement; and

- reduced manning, since specialists would have broader expertiseand, therefore, fewer specialties (and specialists) would beinvolved in maintaining complex systems.

The Electronic Principles Inventory (EPI) was used to gather data aboutthe underlying principles and knowledge required by journeymen in eachspecialty. The EPI asks 1257 questions that can be answered "yes" or "no."The questions pertain to what the technician does on the job and the elec-tronic knowledge used to perform the job. The sample questions shown in thereport ask whether the specialist works with various types of circuits.

The data were gathered from 12,295 job incumbents as a part of the AirForce Occupational Survey Program. The respondents were five-skill-levelspecialists. Twenty-three specialties and two career fields were involved.Seventeen specialties were part of the Communications-Electronics careerfield, and six were in the Wire-Communications career field.

Cluster analyses, correlational analyses, and analyses of variance wereperformed to determine which specialties within each of the two groups had themost similar skill/knowledge requirements. If the specialties were regroupedin accordance with the study recommendations, the original 23 specialtieswould be combined into 9 specialties. Two of the 23 specialties had such lowrequirements for electronic skill/knowledge that they were not recommended forconsolidation with any other(s).

The EPI could be helpful for the CAMT analysis of electronic maintenancetasks. The EPI could help identify the "behavior objects" of each task. Itasks questions like: Do you work with coupling devices in your present job?If y s, do you work with any of the following types of coupling circuits?

- Directly coupled circuits- Capacitive-resistive coupled circuits- Capacitive-inductive coupled circuits- Transformer coupled circuits

3 - 17

Of course, the EPI covers only electronic principles.

Shannon, R. H. and Carter, R. C. (1981, September). Task Analysis and theAbility Requirements of Tasks: Collected Papers (NBDL-81R009). NewOrleans: Naval Biodynamics Laboratory. (AD Al11181)

This is a collection of five papers presented at various professionalmeetings during 1980-1981. They discuss using task analysis to identify theabilities required for jobs as a first step in designing performance tests.

The first four papers describe task analytic methods used to classifyrecurring naval student-pilot flight errors during primary flight training inthe T-34 aircraft. The fifth paper described the application of McCormick'sPAQ to two Navy work stations that have similar tactical missions butdifferent environments. The PAQ was applied to (1) the tactical tasks of theCombat Information Center of Navy ships and (2) the tasks of the CombatInformation Control Officer aboard the E-2 aircraft.

The T-34 student-pilot studies used the following taxonomic outline:

A. Continuous Operations - Tasks involving multidirectional trackingresponses.

1. Pitch Axis Control - Aircraft control in nose up/down axis.

Maintain (a) altitude, (b) airspeed, (c) nose attitude, (d) stickpressure.

2. Roll Axis Control - Aircraft control in the wing up/down axis.

Maintain (a) angle of break, (b) distance, (c) heading, (d) rateof descent, (e) stick pressure.

3. Yaw Axis Control - Aircraft control in the nose left/right axis.

Maintain (a) balanced flight, (b) heading, (c) rudder pressure.

4. Thrust Axis Control - Aircraft control of forward movement.

Maintain (a) rate of descent, (b) throttle pressure.

5. Brake Control - Aircraft control during ground operations byturning, stopping, or changing speed.

Maintain (a) brake control.

3 - 18

B. Discrete Operations - Tasks involving individual distinct movementsor mediating responses.

1. Procedural - Control of aircraft subsystems by not omitting,reordering, or improperly performing necessary sequential steps.

Use (a) flaps, (b) landing gear, (c) fuel switches/lever,(d) canopy, (e) throttle, (f) prop, (g) battery/magnetos,(h) equipment, (i) checklists.

2. Anticipation/Planning - Tasks involving judgment, planning, and"being ahead" of the aircraft.

Anticipate (a) aircraft, (b) position, (c) altitude,(d) airspeed; determine (e) wind direction, (f) landing site(g) location.

3. Communication - Tasks involving transfer of information from one

source to another.

Communicate (a) verbally, (b) visually.

4. Monitor - Searching and scanning inside/outside the cockpit foraircraft safety and maintenance of flight.

Scan (a) for aircraft/obstructions, (b) temperature/pressureinstruments.

Some of the definitions that were advanced are relevant to the presenteffort.

Discrete operations are defined as individually distinctmovements or mediating responses elicited by environmental cues.Continuous operations contain those tasks involvingmultidimensional tracking responses to either contact cues outsidethe cockpit or flight instrument cues within the cockpit. A taskactivity is a qualitative category whose main function involveseither a sensory, cognition, motor, or coordinated perceptual-motortask... Finally, functional objectives are tasks having the sameactivity, goal orientation, sensory cues, and task primitives. Forexample, continuous pitch axis control contains the objectives tomaintain altitude airspeed, nose attitude, and stick pressure(p. 1).

This study pertained to operator tasks and not maintenance, as does CAMT.

Shriver, E. L., Fink, C. D., and Trexler, R. C. (1961, July). A ProceduralGuide for Technical Implementation of the FORECAST Methods of Task andSkill Analysis. Washington, DC: Human Resources Research Office,Training Methods Division. (AD 262771)

3 - 19

Task FORECAST is said to be an attempt to provide methods for improvingthe accuracy of forecasting the skills and knowledges that electronicmaintenance personnel must have to maintain their systems. The FORECASTapproach is to have experts organize the system by developing schematic andblock diagrams, and to determine the training demands from such informationabout the system. In actuality, FORECAST seems to be much more concerned withgiving electronic technicians good performance aids (i.e., schematics) thanwith "forecasting" training needs. This report has no relevance to CAMT. The"Guide" fails to cover the topic of how to derive training needs. However, itis true that when the task is better defined, through good performance aids,it is easier to identify the information that must be covered in training.Applying the FORECAST approach would either reduce the length of electronicstraining or increase the effectiveness of the existing hours devoted to suchtraining.

This manual is concerned primarily with troubleshooting.

Troubleshooting is defined as "identifying the cause of anout-of-tolerance system output. In electronic equipment it is aprocess which involves the successive elimination, by interpretationof symptoms and measurements, of those parts of the system that arenot causing the trouble. Using the electronics information at hisdisposal (e.g., signal flow), the repairman makes a series ofdeductions which progressively narrow the source of the malfunctionto one or more out-of-tolerance parts (e.g., resistor, capacitor,cable). Replacement or adjustment of these parts constitutes repairof the system. (p. 5).

A repairman should have: (1) sufficient knowledge to compute the correctvalue at every possible check point in the system and, therefore, (2) theknowledge to determine the parts of the system that affect the values at everypoint. This means he needs two primary types of electronics knowledge: basicelectronics and system-specific electronics. Basic electronics deals withgeneral methods for computing circuit values. System-specific information isconcerned with detailed circuit analyses which describe the effects producedby those circuits. In addition, repairmen are also provided with someinformation on the probabilities that various types of parts will malfunction.They use this knowledge to determine which parts in the system are withintolerance and which are out of tolerance for any particular malfunction. Thisparagraph is paraphrased from page 6 of the report. It is the closest thereport comes to discussing the derivation of skill and comprehensionrequirements.

The manual devotes considerable attention to "blocking" an electronicsystem.

A troubleshooting block consists of a fairly small group ofparts which has one or more well-defined inputs and outputs. Therelation of the block to its parts is such that, when all blockinputs are good and one or more block outputs are bad, the malfunc-tion(s) must, in all probability, be produced by one or more of theparts located within the block. Troubleshooting blocks are

3 - 20

conceptually similar to 'functional stages' in an electronic systembut are different in that the relation of the block parts to theblock outputs has been rigorously defined. In addition, atroubleshooting block may sometimes contain more than one stage" (p.14). Taking measurements at block inputs and outputs enables thetechnician to rapidly narrow the source of the malfunction(s).

Siegel, A. I., Federman, P. J., and Welsand, E. H. (1980, December).Perceptual/Psychomotor Requirements Basic to Performance in 35 AFSpecialties (AFHRL-TR-80-26). Brooks AFB, TX: Air Force Human ResourcesLaboratory, Manpower and Personnel Division. (AD A093981)

After reviewing the literature pertaining to taxonomies, measurementconsiderations, and job analyses, a taxonomy with 13 perceptual/psychomotorclasses was devised based on Fleishman's work. "These final 13 abilities andtheir respective definitions are as follows:

1. Control Precision--the ability to perform rapid, precise, fine-controlled adjustments by either arm and hand movements or legmovements.

2. Manual Dexterity--the ability to perform skillful,well-directed arm and hand movements to manipulate eitherfairly large or fairly small objects.

3. Finger Dexterity--the ability to perform skillful manipulationsof small objects with the fingers.

4. Multil'mb Coordination--the ability to coordinate the movementsof a number of limbs simultaneously, e.g., two hands, two feet,and hands and feet together.

5. Rate Control (Tracking)--the ability to perform continuousanticipatory motor adjustments relative to changes in speed anddirection of a continuously moving object.

6. Visual Speed and Accuracy--the ability to perceive smalldetails quickly and accurately.

7. Visual Memory--the ability to recall and state verbally orrecall and reproduce through writing and drawings based on pastvisual experiences.

8. Position Memory--the ability to recall rapidly and accuratelythe position of objects from past experience.

9. Auditory Discrimination--the ability to discriminate andinterpret sounds.

3 - 21

10. Auditory Memory--the ability to recognize and reproduce eitherverbally or in writing prior auditory experiences.

11. Clerical Perception-the ability to read or copy rapidly andaccurately pertinent details in scales, graphs, or charts.

12. Perception of Size and Form--the ability to see slightdifferences in the size and shape of objects.

13. Depth Perception--the ability to determine the position ofobjects in space and to perceive in three dimensions"(pp. 26-29).

The methods were tested in two career fields and were applied in alarge-scale data-acquisition effort which included 35 career fields. Thiswork involved over 800 job incumbents at 10 Air Force bases. A factor analy-sis of the data indicated that the perceptual/psychomotor ability taxonomy canbe described by three factors: visual, auditory, and manual factors.

The taxonomy developed by Siegel et al. is based on anability-requirements approach to classification. CAMT task primitives will bebased on behaviors instead of abilities. Siegel's taxonomy, however, maysuggest basic-skill requirements, associated with specific primitives, whichwould be listed in the CAMT Task Primitive Dictionary.

Training Developments Institute (1979, August). Job and Task AnalysisHandbook (TRADOC Pam No. 351-4). Fort Monroe, VA: Headquarters, UnitedStates Army Training and Doctrine Command.

This handbook describes how task analysis should be done in the Army. Itincludes a Task Analysis Worksheet (TRADOC Form 550). The major items ofinformation to be gathered on this worksheet are the following:

1. Task title. - Begins with an action verb.

2. Task number.

3. Task conditions. - Tools, equipment, facilities, environment, etc.

4. Standard. - Procedures to be followed, time limit, errors permitted,production rate, tolerances, etc.

5. Job title. - Duty position.

6. Job aid recommended. - "Yes" or "no" and type.

7. Supervisory job? - "Yes" or "no."

8. Hazard potential. - During training and during the job performance.

3 - 22

9. Safety certification requirements. - Whether certification is

required and the agency that issues it.

10. Equipment used to perform task.

11. Task selection data for training. - These data are derived from aquestionnaire administered to job incumbents. One method forselecting training tasks involves asking the eight questions listedbelow about each task.

Following the data categories is an indication of how theinformation is obtained from job incumbents.

a) Percent performing. - "Do you perform this task?"

b) Time between training and task performance. - 7-point scale. (1)Not yet performed; (2) more than 4 years; (3) 2 to 4 years; (4) 1to 2 years; (5) 6 months to 1 year; (6) 3 to 6 months; (7) duringfirst 3 months of assignment.

c) Frequency of performance. - 4-point scale. (1) Never perform;(2) less than once per month; (3) at least monthly, but less thantwice per week; (4) twice per week or more.

d) Time spent performing task. - 7-point scale. 1 = very much belowthe average time spent on other tasks of the job; 7 = very muchabove the average time spent on other tasks of the job.

e) Consequences of inadequate performance. - 7-point scale. 1 -

negligible consequences; 7 = may result in mission failure,injury, death, or damage to important equipment.

f) Probability of inadequate performance (in comparison with othertasks of the job). - 5-point scale. 1 = rarely if ever; 3 -

about as often as other tasks; 5 = very often.

g) Task delay tolerance. - 7-point scale. 1 = performance can beput off indefinitely; almost never urgent. 7 = must begininstantly.

h) Task learning difficulty. - 7-point scale. From very easy (1) tovery difficult (7).

12. References.

a) Used in analysis.

b) Required to accomplish task.

13. Current training materials.

3 - 23

14. Enabling skills and knowledges required.

a) Baseline entry level. - Normal-repertoire behaviors required bythe task.

b) Skill hierarchy. - This is an inverted-tree diagram that showsthe training objectives associated with the task and the enablingobjectives for each higher-level objective. The skills andknowledges for each training objective are listed. TheHandbook's guidance is probably not sufficient to enable noviceanalysts to construct such diagrams. It is somewhat confusing.

15. Performance elements/steps. - Each step is listed, and four items of

information are recorded for each step.

a) Step description.

b) Cues. - what stimuli prompt initiation and termination of eachstep?

c) Conditions. -Tools, test equipment, forms, references, etc.

d) Standards. - How can step performance be evaluated?

e) Skills/knowledge. - The skills are called "abilities" (e.g.,"ability to read meters"). An example of knowledge is:"knowledge of polarity." They include normal-repertoirebehaviors as well as behaviors to be included in training.

The methods described in this handbook are directly relevant to CAMT.They are similar to CAMT in that the lowest level of task analysis could bebelow the task level and above the step level. The handbook talks about"performance elements or steps necessary to perform the task." The analyst ispermitted to stop short of the step level of description if using the steplevel would make the task description insignificant, vague, or trivial.Insignificant, vague, or trivial means a level at which all members of thetarget population can perform all of the steps without training. As anexample of "element," the handbook breaks down the task of repair carburetorinto three: clean internal parts, replace worn parts, and adjust mixturejets. CAMT would probably analyze this task into the same three taskprimitives.

Interestingly, the level of description this handbook tries to avoid isprecisely the level of description the ETAP method of Reigluth et al. (1980)strives to achieve. ETAP carries its step and substep analysis to the levelat which each step is described in terms of skills that the new job Incumbentbrings to the job. The Job and Task Analysis Handbook instructs the analystto avoid this level of description, calling it "insignificant, vague, ortrivial." CAMT avoids this very detailed level of description because it isnot necessary for the MPT&S decisions that need to be made.

3 - 24

Wheaton, G. R., Rose, A. M., and Fingerman, P. W. (1975, September). Methodsfor Predicting Job-Ability Requirements IV: Task Characteristics,Ability Requirements, and Problem-Solving Strategies (R75-2).Washington, DC: American Institutes for Research. (AD A015719)

This report describes the fourth and final study in a program of researchdealing with the relationships between the characteristics of human tasks andthe abilities required for task performance. The previous studies found thatcomplex changes in the ability requirements related to performance occurred inresponse to variations in task characteristics. In the fourth study, possibleinteractions among task variations, ability profiles, and subject strategieswere examined within the context of the troubleshooting and problem-solvingtasks previously studied. In general, knowledge of a subject'sproblem-solving strategy was useful in obtaining a clearer understanding ofability requirements under different conditions of task performance.

Wilson, M. G., Faucheux, G. N., Gray, J., Wilson, E: B., Lamb, T. A., andGeorge, J. L. (1988, May). Optimizing Aircraft Maintenance Task/Specialty Allocations (AFHRL-TP-87-46). Brooks AFB, TX: Air Force HumanResources Laboratory, Logistics and Human Factors Division. (AD B122183)

This report describes Small Unit Maintenance Manpower Analysis (SUMMA),which uses task analysis and MPT data to decide how best to consolidatemaintenance specialties to meet predetermined operational and maintenancescenarios. Specifically, the SUMMA project was designed to look at threequestions: How can we disperse from larger fixed bases to smaller deployedsites? How many maintenance technicians will we need? Can we do the jobwithout a large increase in the number of technicians?

The task analysis gathered the following information about each task.

From existing data sources (such as the technical orders, the CentralizedData System, and the Mission Essential Subsystem List), data were gathered onthe following nine characteristics:

- task criticality,

- work unit code,

- mean time between occurrences,

- preconditions for task initiation,

- safety precautions,

- follow-on tasks,

- repeat discrepancies,

- type of maintenance (scheduled vs. unscheduled), and

3 - 25

- other tasks that cannot be performed simultaneously.

O-level SMEs were questioned regarding the following topics.

1. Level of Difficulty (under ideal conditions). Judged on afive-point adjectival scale.

2. Reason for Task Difficulty. Reason(s) for a task to be judgedfour or five on the difficulty scale.

3. How Skill Is Acquired. Major source from which respondentsacquired proficiency in each task.

4. How Long This Skill Can Be Retained Without Practice. For eachtask, in months.

5. Number of Repetitions Required to Reach Proficiency. Number ofrepetitions before the respondent was certified as qualified.

6. Crew Size. Minimum number of persons who can safely performthe task.

7. AFSs That Normally Assist. If assistance is required, AFS andskill level.

8. Clock Time Required. Clock time to perform the task.

9. Probability of Successful Completion. Probability, in percent,that the task will be performed without error (including errorsfound through inspection and verification).

10. Suggested Alternate AFSs for Task Performance. Other AFSs thatcould easily be trained to perform the task (in order ofdesirability).

11. Clock Time Required (Alternate AFSs). For each alternate AFS,an estimate of the clock time required for performance.

12. Probability of Satisfactory Completion (Alternate AFSs). Foreach alternate AFS, an estimate of the probability ofsatisfactory task completion.

13. Electronics Knowledge/Ability Required. Knowledge and abilityto make practical applications of electronics principles,concepts, and devices.

14. Mechanical Knowledge/Ability Required. Knowledge and abilityto make practical applications of mechanical principles,concepts, and devices.

3 - 26

15. Fluids and Gases Knowledge/Ability Required. Knowledge andability to make practical applications of the principlesgoverning the behavior of fluids and gases.

16. Microprocessor and Computer Knowledge/Ability Required.Knowledge and ability to make practical applications of theprinciples and concepts involved in digital computing devices.

17. Metal Working Knowledge/Ability Required. Knowledge andability to work with ferrous and non-ferrous metals in therepair, fabrication, and care of metal surfaces, objects, anddevices.

18. Knowledge of Aircraft Structure/Systems Required. The extentto which general knowledge of aircraft structure and thelocations of components, wires, tubing, etc. are required.

19. Muscular Effort. Strength and/or stamina required.

20. Adherence to Procedures. The degree to which fixed or setprocedures are required.

21. Number of Procedural Steps. The number of individual steps orresponses required.

22. Decision Making/Problem Solviny. The extent to which decisionmaking or problem solving is required, as opposed to setprocedures.

23. Working with Hazardous Procedures/Materials. The extent towhich hazardous procedures must be performed or hazardousmaterials must be dealt with.

24. Reading/Using Complex Instructions. The extent to whichcomplex written instructions, diagrams, or other printedmaterials are required.

The following five questions were added for I-level SMEs.

1. After task proficiency is lost, how many uses are required toretrain? The number of times the test station must be used inperforming a maintenance task to regain proficiency.

2. Can this skill be taught entirely on the job to someone fromthe Avionics (32XXX) career field? Could someone from the32XXX career field (other than 326X4) learn this skill entirelyby the OJT method; i.e., without the benefit of additionalresident school training?

3. Could someone of average electronics aptitude (40-50) betrained to perform this task? Yes or no.

3 - 27

4. How frequently do you use the following test stations? Thefour AIS test stations are rated on the following scale:daily, weekly, monthly, every six months, or once a year.

5. If this test station is totally out of commission, are therealternate procedures that could be used to accomplish all orsome of the test objectives? If "yes," list up to threealternate procedures and the testing objectives that aresacrificed by using each one.

This report discusses the MPT process, its decisions, and the job factorsto be considered in making those decisions. It can help the development ofCAMT by suggesting variables that can be used to populate the Task PrimitiveData Base and the Basic Skill and Comprehension Requirements Data Base.

Summary of Literature Review

This review of civilian and military literature has revealed that theCAMT task primitive approach has not previously been used. None of theresearch examined had as its aim the construction of a data base which couldbe used repeatedly in making MPT&S decisions in the design of new weaponsystems.

Several studies focused on what humans can do (abilities); CAMT focuseson what humans actually do--on "task primitives" derived from interviews withworkers at the job site and observation of the task environment. Thesestudies included: Fleishman and Quaintance, 1984; Driskill et al., 1989;Primoff, 1988; Siegel et al., 1980; and Wheaton et al., 1975. If theseabilities proved to be prerequisites for task primitive performance, theycould be listed as basic skill requirements in the Task Primitive Dictionaryand documented during CAMT analysis as personnel requirements.

Only one of the documents analyzed jobs and tasks at the level ofdescription we consider appropriate for facilitating weapon system design.This was the "Job and Task Analysis Handbook" written by the Army's TrainingDevelopment Institute (1979). The most detailed level of task analysis wasunquestionably used by Reigluth et al. (1980). Some of their methods foridentifying comprehension and skill requirements might be adapted for use inCAMT.

The following reports provided clues as to how CAMT task primitives couldbe defined or how the CAMT data base might be populated.

Driskill et al. propose a new AF taxonomy of occupationalabilities that cross occupational AFSs. Common abilitiesrequirements would be derived by analyzing a cluster of taskstatements containing the same verbs. If implemented, the choiceof anchoring verbs might also serve to define task primitives.Relevant abilities requirements could then be transfered from thetaxonomy into the CAMT Task Primitive Dictionary.

3 - 28

- Hosek et al. report that TPDC is attempting to catalogue theinitial skill training of airmen. If TPDC collects coursedesciptive data and course content data--including the time spenton each topic--into an Initial Skill Training Data Base, thisdata could also serve to populate the CAMT Comprehension andBasic Skill Requirements Dictionary.

- Perrin et al. describe how the TDS is clustering maintenancetasks into Task Training Modules TTMs using co-performancecriteria. If these TTMs could form a basis for task primitivedefinition, the TDS data base could populate much of the CAMTTask Primitive Dictionary.

- Ramsey-Klee's description of NOTAP referenced the program'sfrequency rank ordering of equipment, tools, systems, andsupplies which incumbents in each of five Navy enlisted ratingsuse for operating or repairing weapon systems. This NOTAP listcould suggest behavior objects of use for defining taskprimitives.

- Ruck's description of the EPI suggests that the electronicprinciples required for avionics maintenance tasks could beadopted as task primitives, or at least as a source of behaviorobjects. If so, the EPI data base could populate much of theCAMT Task Primitive Dictionary for these tasks.

- The Training Development Institute's approach to job and taskanalysis (TRADOC Pamphlet 351-4), if used, would provide much ofthe data needed in a CAMT Task Primitive Dictionary. Becausethey were published in 1979, there has been ample time to gainexperience with these procedures. Their resultant data basemight be accessible immediately. However, as yet we have noknowledge of how extensively these procedures have been used.

The next CAMT study will explore in greater depth those research programsat TPDC, the Armstrong Laboratory divisions at Brooks AFB, and the the AirForce Occupational Measurement Squadron.

3 - 29

SECTION IV. SUMMARY AND CONCLUSIONS

1. The CAMT concept responds to the need for a cultural translation betweenhuman engineering design for maintainability and its consequent influenceon MPT&S requirements. Specificly, it addresses a system engineeringscenario in which a human engineering analyst evaluates proposed subsystemdesign options in terms of their estimated typical MPT&S resourcerequirements for all related maintenance tasks. Additionally, CAMTprovides immediate human engineering feedback to the subsystem designerfrom indexed standards and lessons learned, and design-related input totraditional MPT&S analyses once the design is chosen.

2. CAMT assumes that because MPT&S analyses typically address entiremaintenance tasks and human engineering analyses typically addressdetailed task steps, analyses that link human engineering with MPT&S mustaddress intermediate-sized portions of maintenance performance: taskprimitives. The concept of a task primitive was refined. Its use as thefoundation for a new type of MPT&S data base was developed.

3. The CAMT data base would contain several designs associated with eachcommon task primitive. Each design, in turn, would be associated withdocumented MPT&S consequences. CAMT would provide, in essence, multiple-baselines with which to compare the proposed design, and the baselines'pieces could be reapplied to other tasks covering other subsystems. Aswith currently practiced comparability analyses, a CAMT would requireexpert judgement to extrapolate human performance from past to futuredesigns. The CAMT data base, however, would significantly narrow thescope of individual judgements to enable more credible extrapolations,especially for radically new designs.

4. A review of civilian and military literature revealed that CAMT's taskprimitive approach has not previously been used. The methodology mostclosely approximating CAMT was found in the "Job and Task AnalysisHandbook" written by the Army's Training Developments Institute (August1979). The handbook addresses "performance elements or steps necessary toperform the task." The analyst is permitted to stop short of the steplevel of description if using the step level would make the taskdescription insignificant, vague, or trivial. "Insignificant, vague ortrivial" means a level at which all members of the target population canperform all of the steps without training. As an example of what is meantby "element," the handbook breaks down the repair carburetor task intothree elements: clean internal parts, replace worn parts, and adjustmixture jets. CAMT would probably analyze this task into the same threetask primitives. The literature review also revealed potential aids todefining task primitives and building the data base around them. We willexplore these in future studies.

5. we conducted a field test of CAMT methodology addressing the feasibilityof defining useful task primitives. We interviewed engine shop crewsupervisors at Dyess AFB concerning three remove-and-install tasks as

4 - 1

performed on three weapon systems. The tasks involved the followingengine components: starter, fuel pump, and turbine wheel assembly. Inaddition to identifying the task steps, behavior objects, constraints, anderrors for each task, we asked the supervisors to construct curricula fortraining engine shop technicians. Finally, we proposed task primitiveswhich we tested, as a set, against criteria of generality, orthogonality,and completeness.

6. Our proposed set of task primitives substantially met these criteria,although many primitives addressed too general a level of behavior objectto readily link to design consequences.

Generality. Task primitives defined for SAC tasks successfullyapplied to MAC tasks as well. Common primitives applied toequivalent tasks for an old jet engine (J-57), a newer jet engine(F-101), and an old turboprop engine (T-56). Common primitives alsoapplied to various tasks on the same engine. Six task primitivesappeared in all nine tasks.

Orthogonality. The originally identified task primitives were notall orthogonal. We deleted four to create an orthogonal list.

Completeness. To test completeness, we compared the trainingcurricula constructed by the interviewees with the task primitivelist. Training curricula topics not covered under the current taskprimitive list would have been in the associated Basic Skill andComprehension Requirements Dictionary, or in task primitives definedto cover a larger task set than our limited sample.

7. A more narrowly scoped set of task primitives could be generated from ourfirst proposed set by defining separate task primitives for more specificbehavior object types. This set would inherit the first set's generality,orthogonality, and completeness; however, it would more readily allow thedata base to store at least three representative samples of each behaviorobject which would typify many designs.

8. We recommend additional research to expand the CAMT feasibility explora-tion to a wider variety of tasks. To supplement the investigation ofmechanical tasks, we should examine electronics tasks. To supplement theinvestigation of shop-level remove-and-replace tasks, we should examineflightline remove-and-replace, inspection/diagnostic, and bench-testingtasks. Subsequent research should expand the exploration of taskprimitive definition from primarily motor tasks to primarily perceptualand cognitive tasks.

9. Further studies must address the broader feasibility issues of efficientdata base population and the reliability/utility of analysis results.

4-2

SECTION V. REFERENCES

Farina, A. J., Jr., and Wheaton, G. R. (1973). Development of a Taxonomy ofHuman Performance: The Task Characteristics Approach to PerformancePrediction (Ms. No. 318). JSAS Catalog of Selected Documents inPsychology, 3, pp. 26-27.

Fleishman, E. A. (1967). Performance Assessment Based on an EmpiricallyDerived Task Taxonomy. Human Factors, 9, pp. 349-366.

Loose, D. R. (1989). Chapter 2, Design for Maintainability Consequences.Design for Maintainability: What Military Standards Do and Don't Say,(AFHRL-TP-89-28), pp. 2-1 - 2-17.

Miller, R. B. (1967). Task Taxonomy: Science of Technology? In W. T.Singleton, R.S. Easterly, and D. C Whitfield (Eds.), The Human Operatorin Complex Systems. London: Taylor & Francis.

Primoff, E.S. (1975). How to Prepare and Conduct Job Element Examinations

(Tech. Study 75-1). Washington, D.C.: Government Printing Office.

Silverman, J. (1967, November). New Techniques in Task Analysis (SRM 68-12).San Diego: U.S. Naval Personnel Research Activity. (AD 663 135)

Theologus, G. C., Romashko, T., and Fleishman, E. A. (1973). Development of aTaxonomy of Human Performance: A Feasibility Study of Ability Dimensionsfor Classifying Human Tasks. JSAS Catalog of Selected Documents inPsychology, 3, pp. 25-26. (Ms. No.321)

5-1

APPENDIX A

BEHAVIOR OBJECTS, CONSTRAINTS, AND ERRORS

FOR THREE TASKS ON THREE AIRCRAFT

A-i

Fuel Pump Removal and Installation - T-56 Engine - C-130 Aircraft

Behavior Objects

High-pressure fuel filter hosesBoltsWashersFuel filterSafety wireSafety-wire pliers1" open-end wrench5/16" universal socket1/4" speed handleRubber fuel containerBracketsCannon plugsSafety wireDuckbill pliersCommon slip-joint pliersFilterB-nutsJamnuts3/8" box-end wrenchDiagonal pliers7/16" socket1/4" ratchet handleSealsFilterAdel clamp3/8" socket9/16 open-end wrenchFuel pumpSpecial fuel-pump-flange wrench (CX1416)Gaskets11/16" open-end wrench1-1/8" socketTorque wrenchOil (MIL-L-6081, grade 1010)Oil (MIL-L-23699)Accessory-gearbox padPutty knifeRags3/8" crowsfoot7/8" crowsfoot1" crowsfoot9/16" crowsfoot

A- 2

Constraints

The top inboard fuel-pump-mounting nut is blocked by the pump, the accessorygear box, and the compressor housing. A special wrench must be used.

When the top inboard fuel-pump-mounting nut is being installed, it can betightened with a special wrench but it cannot be torqued.

The engine must sometimes be turned by hand to align the splines.

Torquing the B-nuts of the fuel filter hoses is difficult due to the presenceof other hoses and lines.

Errors

Failure to properly drain, contain, and dispose of residual fuel could causeinjury to personnel and environmental damage.

Improper safety wiring could cause components to loosen.

Dropping a nut into the belly pan of the engine could cause foreign object

damage.

Failure to install the bracket correctly could cause chafing.

Failure to remove all of the old gasket could cause a fuel leak.

Pinching the seal when installing the inlet line could cause a fuel leak.

Cross-threading or overtorquing B-nuts can shear the threads and cause a fuelleak.

A - 3

Starter Removal and Installation - T-56 Engine - C-130 Aircraft

Behavior Objects

StarterPressure sensing lineDrain plugInlet air ductElectrical leadsEngine gear caseTorque-meter-pickup harnessB-nuts1/2" open-end wrenchMarman clampClamp nut7/16" deep-well socket3/8" ratchet handleSpecial EXQ-31 starter wrench (9/16"; 3/8" drive)3/8" breaker bar12" extension9/16" locknuts and washersGasketPutty knife9/16" box-end wrenchRubber oil containerOil (MIL-L-23699)Safety wireSafety-wire pliersEngine drive-gear female splinesStarter drive-shaft male splinesRagStiff-bristle, nonmetallic brushCleaning solvent (P-D-680, Type II)Paste lubricant (DoD-L-25681)Torque wrench1/2" crowsfoot9/16" crowsfootSyringe

Constraints

The oil pump and the torque meter obstruct access to the locknuts that securethe starter to the engine drive case.

The generator blast tube duct and various electrical leads obstruct removal ofthe starter.

The weiaht of the starter necessitates two technicians for removal andinstaliation.

A- 4

Errors

Nuts or washers dropped into the engine compartment could cause Foreign ObjectDamage (FOD).

Dropping the starter could injure personnel or damage equipment.

Failure to remove all of the gasket could result in an oil leak after

installation.

Prolonged physical contact with oil could be harmful to personnel.

Failure to clean and lubricate splines during installation could cause an oilleak, heat build-up, and seizing.

Undertorquing could result in leaks.

Overtorquing or cross-threading could shear off threads.

Overfilling the oil could cause the seal to fail and leak.

Underfilling the oil could cause the starter to seize.

Improper safety wiring could cause a fastener to loosen.

A - 5

Compressor Turbine Removal and Installation - T-56 Engine - C-130 Aircraft

Behavior Objects

Self-locking nutSafety wireDiagonal cutting pliersInner rear exhaust coneSpecial cone puller (No. 6795867)3/8" socket3/8" ratchet handleThermal insulation blanketBoltsSafety-wire pliers5/16" socketSpeed handleRubber malletMetallic O-ring sealGasketEngine oilLockringTie boltTie-bolt locknutSpecial tie-bolt locknut wrench (No. 6796530)Special Y-bar support (No. 6796382)Rear scavenge pump drive couplingTie-bolt lockpinSpecial turbine-to-compressor tie-bolt spanner wrench (No. 6796533)Heavy malletFire seal collar access doorsCommon screwdriverCam-lock screwsIgniter plugsLiner supports1/4" ratchet handle12" extensionThermocouple wiring harness1/4" speed handle3/8" box-end wrenchNuts3/4" box-end wrenchLifting adapterHoistSpecial turbine coupling shaft holding fixture (No. 6796621)Turbine unitTurbine transport stand3/8" box-end crowsfootDial indicator (total-travel gauge)Axial-clearance gauge adapterOffset screwdriver, comon

A - 6

7/16" box-end wrenchPencil and paperSpecial turbine rotor axial movement positioning wrenchTorque wrenchBreaker barHeat-treated safety wire

Constraints

The safety wire holding the inner-rear-exhaust-cone retaining nut is difficultto remove completely because it is in a figure-eight configuration and passestwice through the hole in the stud.

High torque is required to break loose the tie-bolt retaining locknut. Thespecial wrench used is a special socket attached to a breaker bar.

The four top bolts of the combustion outer casing-to-turbine inlet casesplitline are difficult to break loose.

When positioning the turbine unit for installation, it is difficult to obtainthe correct alignment to engage the splines on the compressor rotor shaft.

The securing screw for the gauge adapter is located inside the support case.Therefore, an offset screwdriver must be used when attaching the dialindicator.

Installation of the turbine rear scavenge pump can be difficult because accessto some of the bolts is blocked by the pump.

After the nut that secures the exhaust cone has been installed, it must besecured with heat-resistant safety wire wrapped in a figure-eight pattern.

Errors

Prolonged physical contact with oil is harmful to personnel.

If not done carefully, breaking loose the tie-bolt retaining locknut couldendanger personnel or damage engine components and tools.

The four top bolts of the combustion outer casing-to-turbine inlet casesplitline could be rounded off during their removal or installation. Theirnuts could also be rounded off.

If the oil tubes are bent when inserting the turbine unit, they will hinderinsertion or have to be replaced.

Failure to measure the front-to-back travel of the turbine rotor willnecessitate removal and reinstallation.

A - 7

Improper torquing techniques could cause fasteners to loosen and damage the

engine.

Failure to zero the dial indicator will result in incorrect measurement.

Improper safety-wiring techniques could cause fasteners to loosen and damagethe engine.

Overtorquing the thermocouple nuts could damage the thermocouple.

A- 8

Fuel Pump Removal and Instc.lation - J-57 Engine - KC-135 Aircraft

Behavior Objects

Fuel inlet tubeWaste fuel containerFuel supply lineSafety wireSafety wire pliersJP-4 fuelFlange nutsFlange boltsWashers1-7/8" open-end wrench7/16" universal socket6" extension12" extension1/4" ratchet handleO-ring sealsDouble cable clamp; bolt and nutFuel pump discharge lineWater injection switch lead3/8" deep-well socket3/8" open-end wrenchSealsAdel clamps; bolts and nutsFord wrenchBypass-to-pump inlet line1/2" deep-well socket1/2" box-end wrenchReturn-to-first-stage pump inlet line7/8" open-end wrenchSeal-drain line5/8" open-end wrenchWater-sensing tubeAccessory-drive suction lineWire bundleB-nuts9/16" open-end wrenchTem erature-sensing lineWater-sensing tubeValve5/16" shallow socketFiber drift (rod)MalletLocking ring; boltFuel pumpAdapter; bolts3/8" ratchet handle2" extension

A- 9

9/16" deep-well socketGasketO-ring sealsElbows; jam nutsAdapter plate and bracketPetrolatumOil (MIL-L-7808)Acid brushTorque wrenchesRagGo/no-go wear gauge (PWA 17286)Plastilube Moly No. 3 greaseFeeler gaugeFlashlightMirror11/16" crowsfoot7/16" crowsfoot9/16" crowsfoot5/8" crowsfoot7/16" torque adapter5/16" universal socket3/8" universal socket1 7/8" crowsfootGrease (MIL-G-21164)

Constraints

When disconnecting the fuel supply line from the pump and the fuel strainer(and when installing it), access to the nuts and bolts is blocked by the waterpump, the forward mounting bracket, and the supply tube itself.

Safety wire is difficult to remove and install during removal and installationof the bypass-to-pump inlet tube because the nuts are close to the union.

There is limited access to the safety wire on the locking-ring bolts whendisconnecting and connecting the locking ring. There is also little room toswing a mallet when using the fiber drift.

Removal and installation of the clamp that holds the inlet-temperature sensingtube and the burner-pressure sense line are blocked by a wire bundle.

Removal and installation of the fuel pump are blocked by the inlet-temperaturesensing tube and the wire bundle.

Safety wire removal and installation is difficult when removing and installingelbows from the fuel pump because the nuts and studs are close to the union.

Limited access causes difficulty in positioning the crowsfoot to torque theelbow-union nuts.

A - 10

Aaccess to the nuts that need to be torqued when attaching the mountingadapter to the fuel pump is blocked by the fuel-pump flange.

When inspecting the installation of the fuel pump, it is difficult to positionfeeler gauge, flashlight, and mirror because of surrounding components.

When installing the second-stage bypass-to-pump inlet line, it is difficult tosafety wire two of the nuts that attach the elbow fitting to the pump becausethey are blocked by the line.

When the fuel supply line is connected to the main fuel strainer, the forwardwater-pump bracket obstructs the torquing and safety wiring of the attachingbolts.

Errors

Prolonged physical contact with fuel could be harmful to personnel.

Because of the crowded placement of components, tubes, and leads and becauseof sharp objects and safety wire, maintenance workers are likely to damagenuts and bolt heads, and to cut their hands.

Failure to support the fuel pump when removing it from the engine, or wheninstalling it, could result in damage to the splined drive gear.

Failure to remove all of the old gasket from the sealing face could cause anoil leak.

Undertorquing could result in leaks.

Overtorquing or cross-threading could shear off threads.

Improper safety wiring could cause a fastener to loosen.

Improper use of the spline wear gauge could result in damage to the pumpdriveshaft and gearbox.

Failure to properly measure the gap between the accessory bracket and thelockring could cause pump failure.

Failure to properly measure the gap between the fuel pump adapter and themounting pad could cause pump failure.

A- 11

Starter Removal and Installation - J-57 Enoine - KC-135 Aircraft

Behavior Objects

Breech capDisconnect handlecouldnon plugscouldnon-plug pliersSafety wireSafety-wire pliersPneumatic ductV-band clampGaskets1/4" ratchet handle7/16" socketDrain and filler plugsOil (MIL-L-7808)7/8" crowsfootTorque wrenchOil containerRagsStarterMounting clampRetaining bolt in center of engine drive couplingTab washerSpring retaining ringsOutput shaft splinesEngine splinesPlastilube Moly No. 3Acid brushLint-free clothLeather or soft plastic mallet

Constraints

The centrifugal-switch electrical connector has limited access because of thegenerator on one side and the hydraulic pump on the other side.

The oil-cooler tab severely restricts rearward movement of the starter duringstarter removal and installation.

The weight of the starter necessitates two workers for removal andinstallation.

The centrifugal-switch electrical connector must be attached and safety wiredbefore the starter is installed. It cannot be installed later, as specifiedby the Job Guide.

A - 12

It is not possible to "tap the entire circumference of the clamp with leatheror soft plastic mallet," as specified by the Job Guide.

Errors

Prolonged physical contact with oil could be harmful to personnel.

It is easy to pull out the safety-wire holes on the electrical connectors.They are light and made of an aluminum alloy.

Failure to support the starter when removing it from the engine, or wheninstalling it, could result in damage to the splined drive gear.

Failure to remove all of the old gasket from the sealing face could cause an

oil leak.

Undertorquing could result in leaks.

Overtorquing or cross-threading could shear off threads.

Improper safety wiring could cause a fastener to loosen.

Failure to detect metal particles in the oil could result in starter failure.

Failure to detect a loose drive shaft could result in starter failure.

Failure to lubricate the drive shaft could result in starter failure.

Failure to seat the clamp could result in starter failure.

Failure to properly install the gasket at the inlet duct could result in airleaks.

A - 13

Compressor Turbine Removal and Installation - J-57 Engine - KC-135 Aircraft

Behavior Objects

Turbine wheelLifting slingTie rodsNutsGuide tool(PWA 8539)Fel-Pro C-200 antiseize compoundInternal and external splinesOil (MIL-L-7808)PetrolatumNo. 4-1/2 bearing and carbon sealAcid brushHoistSafety hookCoupling wrench adapter (PWA 7317)Coupling wrench collar (PWA 7489)Torque adapter (PWA 7392)Coupling wrench (PWA 6715)Anchor-plate barCoupling wrench barAnchor plateBreaker barTorque wrenchHot-section pencil

Constraints

None

Errors

Failure to attach the lifting sling with the high mark in the proper positionwill result in the turbine wheel being out of balance.

Failure to install the guide tool properly could result in damage to theNo. 4-1/2 bearing.

Failure to lubricate properly could result in equipment damage duringinstallation and seizing during operation.

It is possible to damage the turbine wheel or injure personnel when using thehoist to install the turbine wheel.

If the turbine wheel is not manipulated carefully when seating it into itsfinal position, carbon seals could be damaged.

A - 14

Failure to rotate the compressor slowly as the turbine shaft is being

installed could damage the No. 4-1/2 bearing or carbon seal.

Prolonged physical contact with oil could be harmful to personnel.

Failure to detect positive engagement of coupling pins could result inturbine-wheel failure.

Failure to withdraw the coupling wrench 1/2" to 3/4" will result in thecoupling pins remaining in unlocked position.

Failure to detect continuous contact of turbine blades with the outer air-sealring will result in turbine-wheel failure.

A - 15

Fuel Pump Removal and Installation - F-101 Engine - B-lB Aircraft

(The Main Engine Control (MEC) and main fuel pump are removed together.)

Behavior Objects

1/4" universal socket1/4" ratchet handle12" extension5/16" universal socketFuel supply lineB-nutsSafety wireSafety wire pliers11/16" open-end wrenchPBI tubecouldnon-plug pliersElectrical connectorsVariable Stator Vane (VSV) feedback cableCotter pinSupport bracketBoltsWashersTorque adapterBracketFuel manifoldSeal9/16" open-end wrench

Compressor bleed air pressure tubeCompressor discharge pressure tube5/8" open-end wrenchAugmenter signal tubeInlet Guide Vane (IGV) servo fuel pressure tube

Main fuel discharge tubeGasketsMain fuel pump

MEC assemblyV-band clamp boltAllen wrench

V-band clampPackingsScribe7/16" dog bone7/16" open-end wrenchJP-4 fuelOil (MIL-L-7808)PackingsTorque wrenchesAllen-head wrench adapter

Dowel pin

A - 16

3/8" ratchet handle6" extensionNon-metallic malletFiber drift11/16" crowsfoot3/4" open-end wrench5/8" crowsfoot7/8" open-end wrench9/16" crowsfoot3/4" crowsfoot3/8" box wrench1/4" box wrenchDiagonal-cut pliersNeedle-nose pliers

Constraints

Loosening or tightening the support tube on the VSV feedback cable isdifficult because the support bracket limits wrench travel.

A special tool must be used to remove or install the upper-right bolt from thefuel manifold. A normal 1/4" box-end or dog bone is too long to fit into theavailable space; it is obstructed by the bracket.

Disconnecting or connecting the augmenter signal tube is difficult becausewrench travel is restricted by the IGV servo fuel pressure tube.

Accessibility is very restricted at the V-band clamp bolt. It is difficult toremove the safety wire and to install it.

The weight of the main fuel pump and MEC assembly requires two workers fortheir removal.

Various tubes obstruct the removal and installation of the main fuel pump andMEC assembly.

It is difficult to disconnect or connect the main fuel pump and the MECassembly because one nut is blocked by a filter housing. A dog bone couldnotbe used on this nut when disconnecting it. A crowsfoot couldnot be used fortorquing it.

To mate the main fuel pump drive shaft to the accessory gearbox, the V-bandclamp must be held out of the way until the two components mate tightly.

It is difficult to tap the V-band clamp at some points on its circumferencebecause there is little room to swing the mallet.

A - 17

It is difficult to torque thye fuel manifold assembly mount bolts because themanifold bracket prevents torquing of the upper-right bolt.

Errors

Prolonged physical contact with fuel could be harmful to personnel.

The main fuel supply tube could be dented during removal.

It is possible to pull out the safety wire hole on the B-nut beforedisconnecting the PBl tube and the bleed-air or discharge pressure tubes.

It is possible to pull out the safety wire hole on the jam nut beforeloosening the support tube on the VSV feedback cable.

It is possible to overspread the V-band clamp when removing or installing itand thereby fatigue its metal. It is also possible to round out the Allenhole in the bolt that tightens the clamp.

Failure to maintain the fuel pump level when inserting it or withdrawing itfrom the accessory gearbox could damage the splines.

Failure to use the two-wrench method-when appropriate to remove andinstall B-nuts (and torque them)-could cause damage to tubes or unions.

Undertorquing could result in leaks.

Overtorquing or cross-threading could shear off threads.

Improper safety wiring could cause a fastener to loosen.

Failure to properly inspect the V-band clamp before installation could resultin clamp failure or fuel leaks.

It is possible to install an electrical connector with a bent pin.

Improper connection of the main engine control lever to the NEC lever arm willresult in binding.

A - 18

Starter Removal and Installation - F-101 Engine - B-IB Aircraft

Behavior Objects

7/8" open-end wrenchCase drain plugPackings7/16" open-end wrenchOil (MIL-L-7808)Oil containerSafety wireSafety wire pliersTorque wrenches7/8" crowsfootElectrical connectorcouldnon-plug pliersOil supply tubeB-nut5/8" open-end wrenchAir turbine starter control valveNutsAir turbine starterV-band coupling7/16" deep-well socket1/4" ratchet handleHardness Critical Procedures (HCP) test equipmentPlastilubemalletFiber drift

Constraints

It is difficult to remove or install the air turbine starter control valvebecause the V-band coupling nut couldnot be removed or installed with asocket. The nut is too close to the starter exhaust duct. An open-end wrenchmust be used, and its movement is restricted.

The weight of the starter necessitates two workers for removal andinstallation.

Prolonged physical contact with oil could be harmful to personnel.

Improper torquing could cause leaks.

Improper safety wiring could cause leaks.

If the starter is not supported during removal and installation, splines couldbe damaged.

A - 19

If he oil supply tube is bent during starter removal or installation, its

B-nut could be hard to reattach.

Overtorquing or cross-threading could shear off threads.

Failure to line up the keyway on the electrical connector when installing itcould result in bent pins.

A - 20

Compressor Turbine Removal and Installation - F-101 Engine - B-1B Aircraft

Behavior Objects

Splines, rabbet, and threads of forward low-pressure turbine shaftLint-free clothOil (MIL-L-7808)Inspection gaugeDepth micrometerPencil and paperLift fixtureHoistCenter-of-gravity fixtureTransport standLow-pressure turbineNutsBolts5/16" box-end wrench1/4" ratchet handle5/16" universal socket5/8" open-end wrench1/2" speed handleOutside micrometerPush/pull fixtureHydraulic cylinderShroud sealsFeeler gaugeNeedle-nose pliersProtector of No. 4 bearing carbon sealCarbon seal and outer race area of No. 4 bearingHeat gunHeat duct and heat-duct thumb screwPilot tubeIce bucket (chilling fixture 3C3320)GlovesDry icePetrolatumSurgical glovesRotor fanMeasurement tubeCalculatorSweeney torque multiplierAnti-seize compound (GP-460)i/2"-drive breaker barRingsSmall inspection mirrorTorque wrench3/4" open-end wrenchBracketsWashers

A - 21

Special ratchet-tool !set (3C3121P02)5/16" offset box-end wrenchFairing screwsApex9/16" dogbone7/16" dogbone9/16" socket7/16" socket3/8" ratchet handle5/16" dogbone5/16" universal socket5/16" box-end wrenchTorque adapterSerrated retaining ringThermal ringCoupling nut

Constraints

Measurements are difficuelt to make because of limited space at the seatingrabbet.

There is limited space to pack petrolatum around the No. 4.

It is difficult to install the low-pressure turbine because it must be alignedby touch. It is impossible to visually align it with keyways.

Installirg outer fairing segments is difficult because there is limited spacebetween the flange and the struts. About half of the 76 nuts are difficult toinstall.

Errors

Failure to clean the inspection gauge or failure to hold it flush against theface of the seating rabbet will result in incorrect measurement.

Improper use of the depth micrometer will result in incorrect measurement.

Failure to set the center-of-gravity fixture will result in damage to theengine and the turbine shaft.

Failure to install the push/pull fixture correctly could result in damage tothe engine or injury to personnel.

Incorrect measurement of the turbine-shroud seals could result in suboptimalengine performance.

Overheating the bearing outer ring area could damage the outer ring or thecarbon seal.

A - 22

Failure to observe safety precautions could result in heat burns or dry-iceburns.

Failure to wear surgical gloves when touching the bearing could result in

bearing failure.

Failure to remove dry-ice buckets carefully could damage the carbon seal.

Misalignment of the hoist could damage shaft housing.

Failure to release hydraulic pressure before removing the push/pull fixturecould injure personnel and damage equipment.

Prolonged physical contact with oil could be harmful to personnel.

Failure to check the seating of the serrated retaining ring and the couplingnut could damage the turbine shaft.

A - 23

APPENDIX B

LIST OF ACRONYMS

B-

LIST OF ACRONYMS

Acrony Definition

ADG Accessory Drive GearboxAF (United States) Air ForceAFB Air Force BaseAFLC Air Force Logistics ComandAFMPC Air Force Military Personnel CenterAFS Air Force SpecialtyAGE Auxiliary Ground EquipmentAL/HR The Human Resources Directorate of Armstrong LaboratoryASA Applied Science Associates Inc.ASVAB Armed Service Vocational Aptitude BatteryATC Air Training CommandATC/T The Technical Training staff of ATCBCS Baseline Comparison SystemCAD Computer-Aided DesignCAMT Comparative Anatomy of Maintenance TasksCM Consistently MappedDoD Department of DefenseDODD DOD DirectiveDPML Deputy Program Manager for LogisticsDSC&Cs Design-Specific Characteristics and ConstraintsDTIC Defense Technical Information CenterEPI Electronic Principles InventoryETAP Extended Task Analysis ProcedureFOD Foreign Object DamageHCP Hardness Critical ProceduresHSI Human-Systems IntegrationI-level Intermediate-level maintenanceIGV Inlet Guide VaneILS Integrated Logistics SupportIST Initial Skill TrainingJEM Job Element MethodJFS Jet Fuel StarterLCOM Logistics Composite ModelLRU Line Replaceable UnitMAC Military Airlift CommandMAJCOM MAJor (Air Force) COMmandMAJCOM/LG MAJCOM logistics staffMAJCOM/XPM MAJCOM planning staffMEC Main Engine ControlMOE Measure Of EffectivenessM ManpowerMPT Manpower, Personnel, and TrainingMPT&S MPT and SafetyNOTAP Navy Occupational Task Analysis ProgramO-level Operational-level maintenanceOJT On-the-Job TrainingOPR Office of Primary Responsibility

B-2

LIST OF ACRNYMS (Continued)

Acronym Definition

P PersonnelPAQ Position Analysis QuestionaireQTRD QuesTech Research DivisionR&M Reliability and MaintainabilityS SafetySAC Strategic Air CommandSME Subject-Matter Expert

SPO System Program OfficeSUMMA Small-Unit Maintenance-Manpower Analysis

T TrainingTAC Tactical Air ConuandTCS Task Characteristics SystemTDS Training Decisions SystemTPDC Training & Performance Data CenterTTM Task Training ModuleVM Variably MappedVSV Variable Stator VaneWSAP Weapon-System Acquisition Process

B-3